ids = concepts.stream().map(TypeUtil::getConceptIds).flatMap(Collection::stream)
48 | .filter(synonymExpansionProvider::accept).collect(toSet());
49 | synonymExpansionProvider.getSynonyms(ids);
50 | }
51 | }
52 |
53 | }
54 |
--------------------------------------------------------------------------------
/src/main/java/edu/cmu/lti/oaqa/baseqa/providers/kb/ConceptSearchProvider.java:
--------------------------------------------------------------------------------
1 | /*
2 | * Open Advancement Question Answering (OAQA) Project Copyright 2016 Carnegie Mellon University
3 | *
4 | * Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
5 | * in compliance with the License. You may obtain a copy of the License at
6 | *
7 | * http://www.apache.org/licenses/LICENSE-2.0
8 | *
9 | * Unless required by applicable law or agreed to in writing, software distributed under the License
10 | * is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
11 | * or implied. See the License for the specific language governing permissions and limitations
12 | * under the License.
13 | */
14 |
15 | package edu.cmu.lti.oaqa.baseqa.providers.kb;
16 |
17 | import edu.cmu.lti.oaqa.type.kb.Concept;
18 | import org.apache.uima.UIMAException;
19 | import org.apache.uima.analysis_engine.AnalysisEngineProcessException;
20 | import org.apache.uima.fit.factory.JCasFactory;
21 | import org.apache.uima.jcas.JCas;
22 | import org.apache.uima.resource.Resource;
23 |
24 | import java.util.List;
25 | import java.util.Optional;
26 |
27 | /**
28 | *
29 | * An implementation of this interface can look up a concept (given by its name) in an ontology, and
30 | * return the most relevant {@link Concept}.
31 | *
32 | *
33 | * In comparison, {@link ConceptProvider} is often used before a {@link ConceptSearchProvider},
34 | * which identifies the concepts from a plain text.
35 | * The concepts identified from {@link ConceptProvider} do not have to be existing entries in any
36 | * ontology, since some {@link ConceptProvider}s make "guess" based on their morphological
37 | * structures.
38 | *
39 | *
40 | * @see ConceptProvider
41 | *
42 | * @author Zi Yang created on 4/4/15
43 | */
44 | public interface ConceptSearchProvider extends Resource {
45 |
46 | default Optional search(String string) throws AnalysisEngineProcessException {
47 | JCas jcas;
48 | try {
49 | jcas = JCasFactory.createJCas();
50 | } catch (UIMAException e) {
51 | throw new AnalysisEngineProcessException(e);
52 | }
53 | return search(jcas, string);
54 | }
55 |
56 | Optional search(JCas jcas, String string) throws AnalysisEngineProcessException;
57 |
58 | default Optional search(JCas jcas, String string, String searchType)
59 | throws AnalysisEngineProcessException {
60 | return search(jcas, string, searchType, 1).stream().findFirst();
61 | }
62 |
63 | List search(JCas jcas, String string, String searchType, int hits)
64 | throws AnalysisEngineProcessException;
65 |
66 | }
67 |
--------------------------------------------------------------------------------
/src/main/java/edu/cmu/lti/oaqa/baseqa/providers/kb/SynonymExpansionProvider.java:
--------------------------------------------------------------------------------
1 | /*
2 | * Open Advancement Question Answering (OAQA) Project Copyright 2016 Carnegie Mellon University
3 | *
4 | * Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
5 | * in compliance with the License. You may obtain a copy of the License at
6 | *
7 | * http://www.apache.org/licenses/LICENSE-2.0
8 | *
9 | * Unless required by applicable law or agreed to in writing, software distributed under the License
10 | * is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
11 | * or implied. See the License for the specific language governing permissions and limitations
12 | * under the License.
13 | */
14 |
15 | package edu.cmu.lti.oaqa.baseqa.providers.kb;
16 |
17 | import java.util.Collection;
18 | import java.util.Map;
19 | import java.util.Set;
20 |
21 | import org.apache.uima.analysis_engine.AnalysisEngineProcessException;
22 | import org.apache.uima.resource.Resource;
23 |
24 | /**
25 | * An implementation of this interface can identify the list of synonyms for a given concept ID.
26 | * It should also override the {@link #accept(String)} method to filter the ID that is supported
27 | * by the particular synonym expansion provider.
28 | *
29 | * @author Zi Yang created on 4/20/15
30 | */
31 | public interface SynonymExpansionProvider extends Resource {
32 |
33 | boolean accept(String id);
34 |
35 | Set getSynonyms(String id) throws AnalysisEngineProcessException;
36 |
37 | Map> getSynonyms(Collection ids)
38 | throws AnalysisEngineProcessException;
39 |
40 | }
41 |
--------------------------------------------------------------------------------
/src/main/java/edu/cmu/lti/oaqa/baseqa/providers/ml/classifiers/FeatureConstructorProvider.java:
--------------------------------------------------------------------------------
1 | /*
2 | * Open Advancement Question Answering (OAQA) Project Copyright 2016 Carnegie Mellon University
3 | *
4 | * Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
5 | * in compliance with the License. You may obtain a copy of the License at
6 | *
7 | * http://www.apache.org/licenses/LICENSE-2.0
8 | *
9 | * Unless required by applicable law or agreed to in writing, software distributed under the License
10 | * is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
11 | * or implied. See the License for the specific language governing permissions and limitations
12 | * under the License.
13 | */
14 |
15 | package edu.cmu.lti.oaqa.baseqa.providers.ml.classifiers;
16 |
17 | import java.util.Map;
18 |
19 | import org.apache.uima.jcas.JCas;
20 | import org.apache.uima.resource.Resource;
21 |
22 | /**
23 | * An interface that creates a feature vector from an input {@link JCas}.
24 | *
25 | * TODO: To be migrated to {@link edu.cmu.lti.oaqa.baseqa.learning_base.Scorer},
26 | * where a candidate is given in the method
27 | * {@link edu.cmu.lti.oaqa.baseqa.learning_base.Scorer#score(JCas, Object)}.
28 | *
29 | * @see edu.cmu.lti.oaqa.baseqa.learning_base.Scorer
30 | *
31 | * @author Zi Yang created on 4/5/15
32 | */
33 | public interface FeatureConstructorProvider extends Resource {
34 |
35 | Map constructFeatures(JCas jcas);
36 |
37 | }
38 |
--------------------------------------------------------------------------------
/src/main/java/edu/cmu/lti/oaqa/baseqa/providers/parser/ParserProvider.java:
--------------------------------------------------------------------------------
1 | /*
2 | * Open Advancement Question Answering (OAQA) Project Copyright 2016 Carnegie Mellon University
3 | *
4 | * Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
5 | * in compliance with the License. You may obtain a copy of the License at
6 | *
7 | * http://www.apache.org/licenses/LICENSE-2.0
8 | *
9 | * Unless required by applicable law or agreed to in writing, software distributed under the License
10 | * is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
11 | * or implied. See the License for the specific language governing permissions and limitations
12 | * under the License.
13 | */
14 |
15 | package edu.cmu.lti.oaqa.baseqa.providers.parser;
16 |
17 | import edu.cmu.lti.oaqa.type.nlp.Token;
18 | import org.apache.uima.jcas.JCas;
19 | import org.apache.uima.resource.Resource;
20 |
21 | import java.util.List;
22 |
23 | /**
24 | * An implementation of this interface should wrap a parser to {@link #tokenize(JCas)},
25 | * {@link #tagPartOfSpeech(JCas)}, {@link #lemmatize(JCas)}, and {@link #parseDependency(JCas)} of
26 | * the document text in a input {@link JCas}.
27 | *
28 | * @author Zi Yang created on 3/12/16
29 | */
30 | public interface ParserProvider extends Resource {
31 |
32 | List tokenize(JCas jcas);
33 |
34 | void tagPartOfSpeech(JCas jcas, List tokens);
35 |
36 | default List tagPartOfSpeech(JCas jcas) {
37 | List tokens = tokenize(jcas);
38 | tagPartOfSpeech(jcas, tokens);
39 | return tokens;
40 | }
41 |
42 | void lemmatize(JCas jcas, List tokens);
43 |
44 | default List lemmatize(JCas jcas) {
45 | List tokens = tokenize(jcas);
46 | tagPartOfSpeech(jcas, tokens);
47 | lemmatize(jcas, tokens);
48 | return tokens;
49 | }
50 |
51 | void parseDependency(JCas jcas, List tokens);
52 |
53 | default List parseDependency(JCas jcas) {
54 | List tokens = tokenize(jcas);
55 | tagPartOfSpeech(jcas, tokens);
56 | lemmatize(jcas, tokens);
57 | parseDependency(jcas, tokens);
58 | return tokens;
59 | }
60 |
61 | }
62 |
--------------------------------------------------------------------------------
/src/main/java/edu/cmu/lti/oaqa/baseqa/question/parse/QuestionParser.java:
--------------------------------------------------------------------------------
1 | /*
2 | * Open Advancement Question Answering (OAQA) Project Copyright 2016 Carnegie Mellon University
3 | *
4 | * Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
5 | * in compliance with the License. You may obtain a copy of the License at
6 | *
7 | * http://www.apache.org/licenses/LICENSE-2.0
8 | *
9 | * Unless required by applicable law or agreed to in writing, software distributed under the License
10 | * is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
11 | * or implied. See the License for the specific language governing permissions and limitations
12 | * under the License.
13 | */
14 |
15 | package edu.cmu.lti.oaqa.baseqa.question.parse;
16 |
17 | import edu.cmu.lti.oaqa.baseqa.providers.parser.ParserProvider;
18 | import edu.cmu.lti.oaqa.baseqa.util.ProviderCache;
19 | import edu.cmu.lti.oaqa.baseqa.util.UimaContextHelper;
20 | import edu.cmu.lti.oaqa.type.nlp.Token;
21 | import org.apache.uima.UimaContext;
22 | import org.apache.uima.analysis_component.JCasAnnotator_ImplBase;
23 | import org.apache.uima.analysis_engine.AnalysisEngineProcessException;
24 | import org.apache.uima.jcas.JCas;
25 | import org.apache.uima.resource.ResourceInitializationException;
26 |
27 | /**
28 | * This {@link JCasAnnotator_ImplBase} uses a {@link ParserProvider} to parse and annotate the
29 | * question in the main view.
30 | *
31 | * @see ParserProvider
32 | *
33 | * @author Zi Yang created on 4/12/15
34 | */
35 | public class QuestionParser extends JCasAnnotator_ImplBase {
36 |
37 | private ParserProvider parserProvider;
38 |
39 | @Override
40 | public void initialize(UimaContext context) throws ResourceInitializationException {
41 | super.initialize(context);
42 | String nlpProviderName = UimaContextHelper.getConfigParameterStringValue(context,
43 | "parser-provider");
44 | parserProvider = ProviderCache.getProvider(nlpProviderName, ParserProvider.class);
45 | }
46 |
47 | @Override
48 | public void process(JCas jcas) throws AnalysisEngineProcessException {
49 | parserProvider.parseDependency(jcas).forEach(Token::addToIndexes);
50 | }
51 |
52 | }
53 |
--------------------------------------------------------------------------------
/src/main/java/edu/cmu/lti/oaqa/bioasq/eval/calculator/AnswerEvalMeasure.java:
--------------------------------------------------------------------------------
1 | /*
2 | * Open Advancement Question Answering (OAQA) Project Copyright 2016 Carnegie Mellon University
3 | *
4 | * Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
5 | * in compliance with the License. You may obtain a copy of the License at
6 | *
7 | * http://www.apache.org/licenses/LICENSE-2.0
8 | *
9 | * Unless required by applicable law or agreed to in writing, software distributed under the License
10 | * is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
11 | * or implied. See the License for the specific language governing permissions and limitations
12 | * under the License.
13 | */
14 |
15 | package edu.cmu.lti.oaqa.bioasq.eval.calculator;
16 |
17 | import edu.cmu.lti.oaqa.baseqa.eval.Measure;
18 |
19 | /**
20 | * Definitions of BioASQ Phase B Factoid, List, and YesNo QA evaluation metrics.
21 | *
22 | * @see AnswerEvalCalculator
23 | *
24 | * @author Zi Yang created on 4/29/15
25 | */
26 | enum AnswerEvalMeasure implements Measure {
27 |
28 | // PER-TOPIC FACTOID QUESTION ANSWER MEASURES
29 | FACTOID_COUNT, FACTOID_STRICT_RETRIEVED, FACTOID_LENIENT_RETRIEVED, FACTOID_RECIPROCAL_RANK,
30 |
31 | // ACCUMULATED FACTOID QUESTION ANSWER MEASURES
32 | FACTOID_STRICT_ACCURACY, FACTOID_LENIENT_ACCURACY, FACTOID_MRR,
33 |
34 | // PER-TOPIC LIST QUESTION ANSWER MEASURES
35 | LIST_COUNT, LIST_PRECISION, LIST_RECALL, LIST_F1,
36 |
37 | // ACCUMULATED LIST QUESTION ANSWER MEASURES
38 | LIST_MEAN_PRECISION, LIST_MEAN_RECALL, LIST_MEAN_F1,
39 |
40 | // PER-TOPIC YESNO QUESTION ANSWER MEASURES
41 | YESNO_CORRECT, YESNO_TRUE_POS, YESNO_TRUE_NEG,
42 |
43 | // ACCUMULATED YESNO QUESTION ANSWER MEASURES
44 | YESNO_COUNT, YESNO_MEAN_ACCURACY, YESNO_MEAN_POS_ACCURACY, YESNO_MEAN_NEG_ACCURACY;
45 |
46 | static {
47 | for (AnswerEvalMeasure measure : values()) {
48 | Measure.name2measure.put(measure.getName(), measure);
49 | }
50 | }
51 |
52 | @Override
53 | public String getName() {
54 | return name();
55 | }
56 |
57 | }
58 |
--------------------------------------------------------------------------------
/src/main/java/edu/cmu/lti/oaqa/bioqa/providers/kb/TmToolConceptProviderException.java:
--------------------------------------------------------------------------------
1 | /*
2 | * Open Advancement Question Answering (OAQA) Project Copyright 2016 Carnegie Mellon University
3 | *
4 | * Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
5 | * in compliance with the License. You may obtain a copy of the License at
6 | *
7 | * http://www.apache.org/licenses/LICENSE-2.0
8 | *
9 | * Unless required by applicable law or agreed to in writing, software distributed under the License
10 | * is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
11 | * or implied. See the License for the specific language governing permissions and limitations
12 | * under the License.
13 | */
14 |
15 | package edu.cmu.lti.oaqa.bioqa.providers.kb;
16 |
17 | import java.util.Collection;
18 |
19 | /**
20 | * This {@link RuntimeException} gives detailed debugging information for
21 | * {@link TmToolConceptProvider} related errors.
22 | *
23 | * @see TmToolConceptProvider
24 | *
25 | * @author Zi Yang created on 3/25/16.
26 | */
27 | class TmToolConceptProviderException extends RuntimeException {
28 |
29 | private TmToolConceptProviderException(String message) {
30 | super(message);
31 | }
32 |
33 | private TmToolConceptProviderException(String message, Throwable e) {
34 | super(message, e);
35 | }
36 |
37 | static TmToolConceptProviderException unequalVolume(String trigger, int sentVolume,
38 | int recvVolume) {
39 | return new TmToolConceptProviderException(
40 | "Unequal volume at " + trigger + ": sent " + sentVolume + " texts and received " +
41 | recvVolume + ".");
42 | }
43 |
44 | static TmToolConceptProviderException unequalTextLength(String trigger, String sentText,
45 | String recvText) {
46 | return new TmToolConceptProviderException(
47 | "Unequal text length at " + trigger + ":\nSent: " + sentText + "\nRecv: " + recvText);
48 | }
49 |
50 | static TmToolConceptProviderException textChanged(String sentText, String recvText,
51 | String trigger) {
52 | return new TmToolConceptProviderException("Error at trigger " + trigger);
53 | }
54 |
55 | static TmToolConceptProviderException unknownException(String trigger, Throwable e) {
56 | return new TmToolConceptProviderException("Error at trigger " + trigger, e);
57 | }
58 |
59 | static TmToolConceptProviderException offsetOutOfBounds(String sentText,
60 | Collection denotations, Throwable e) {
61 | return new TmToolConceptProviderException(
62 | "Offset out of bounds:\nSent: " + sentText + "\nSpan: " + denotations);
63 | }
64 |
65 | }
66 |
--------------------------------------------------------------------------------
/src/main/java/edu/cmu/lti/oaqa/bioqa/providers/query/PubMedQueryStringConstructor.java:
--------------------------------------------------------------------------------
1 | /*
2 | * Open Advancement Question Answering (OAQA) Project Copyright 2016 Carnegie Mellon University
3 | *
4 | * Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
5 | * in compliance with the License. You may obtain a copy of the License at
6 | *
7 | * http://www.apache.org/licenses/LICENSE-2.0
8 | *
9 | * Unless required by applicable law or agreed to in writing, software distributed under the License
10 | * is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
11 | * or implied. See the License for the specific language governing permissions and limitations
12 | * under the License.
13 | */
14 |
15 | package edu.cmu.lti.oaqa.bioqa.providers.query;
16 |
17 | import com.google.common.base.Strings;
18 | import edu.cmu.lti.oaqa.baseqa.providers.query.BooleanBagOfPhraseQueryStringConstructor;
19 |
20 | import java.util.Collection;
21 |
22 | /**
23 | * This {@link edu.cmu.lti.oaqa.baseqa.providers.query.QueryStringConstructor} extends the
24 | * {@link BooleanBagOfPhraseQueryStringConstructor} to allow to include field information within
25 | * pairs of square brackets for PubMed
26 | * search.
27 | *
28 | * @see BooleanBagOfPhraseQueryStringConstructor
29 | *
30 | * @author Zi Yang created on 10/8/14
31 | */
32 | public class PubMedQueryStringConstructor extends BooleanBagOfPhraseQueryStringConstructor {
33 |
34 | public static final String FIELD_PREFIX = "[";
35 |
36 | public static final String FIELD_SUFFIX = "]";
37 |
38 | public PubMedQueryStringConstructor() {
39 | super();
40 | }
41 |
42 | @Override
43 | public String formatQueryConcept(String formatQueryField, String formatQueryText) {
44 | return formatQueryText + Strings.nullToEmpty(formatQueryField);
45 | }
46 |
47 | @Override
48 | public String formatQueryField(Collection namedEntityTypes, String conceptType) {
49 | return namedEntityTypes.isEmpty() ? null
50 | : (FIELD_PREFIX + namedEntityTypes.stream().findFirst().get() + FIELD_SUFFIX);
51 | }
52 |
53 | }
54 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/abstract_query/bag-of-token.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.abstract_query.BagOfTokenAbstractQueryGenerator
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/abstract_query/concept-required.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.abstract_query.ConceptAbstractQueryGenerator
2 |
3 | use-weight: false
4 | use-type: false
5 | required: true
--------------------------------------------------------------------------------
/src/main/resources/baseqa/abstract_query/concept.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.abstract_query.ConceptAbstractQueryGenerator
2 |
3 | use-weight: false
4 | use-type: false
5 | required: false
--------------------------------------------------------------------------------
/src/main/resources/baseqa/abstract_query/token-concept.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.abstract_query.TokenConceptAbstractQueryGenerator
2 |
3 | stoplist-path: /dictionaries/stoplist.txt
4 | use-weight: false
5 | use-type: false
6 | #pos-tags-path: /dictionaries/pos-tags.txt
7 | #noun-tags-path: /dictionaries/noun-tags.txt
--------------------------------------------------------------------------------
/src/main/resources/baseqa/abstract_query/token-selection-pos-stoplist.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.abstract_query.token-selection-pos
2 |
3 | stoplist-path: /dictionaries/stoplist.txt
4 | pos-tags-path: /dictionaries/pos-tags.txt
5 | noun-tags-path: /dictionaries/noun-tags.txt
--------------------------------------------------------------------------------
/src/main/resources/baseqa/abstract_query/token-selection-pos.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.abstract_query.token-selection
2 |
3 | # explanations see https://www.ling.upenn.edu/courses/Fall_2003/ling001/penn_treebank_pos.html
4 | pos-tags-path: /dictionaries/pos-tags.txt
5 | noun-tags-path: /dictionaries/noun-tags.txt
6 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/abstract_query/token-selection.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.abstract_query.TokenSelectionAbstractQueryGenerator
2 |
3 | # add parameter values for pos-tags-path and stoplist-path
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/collective_score/classifier-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.learning_base.classifier-predict
2 |
3 | candidate-provider: 'inherit: baseqa.answer.score.candidate-provider'
4 | scorers: |
5 | - inherit: baseqa.answer.collective_score.scorers.original
6 | - inherit: baseqa.answer.collective_score.scorers.distance
7 | - inherit: baseqa.answer.collective_score.scorers.edit-distance
8 | - inherit: baseqa.answer.collective_score.scorers.type-coercion
9 | - inherit: baseqa.answer.collective_score.scorers.shape-distance
10 | # add "classifier"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/collective_score/classifier-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.learning_base.classifier-train
2 |
3 | candidate-provider: 'inherit: baseqa.answer.score.candidate-provider'
4 | scorers: |
5 | - inherit: baseqa.answer.collective_score.scorers.original
6 | - inherit: baseqa.answer.collective_score.scorers.distance
7 | - inherit: baseqa.answer.collective_score.scorers.edit-distance
8 | - inherit: baseqa.answer.collective_score.scorers.type-coercion
9 | - inherit: baseqa.answer.collective_score.scorers.shape-distance
10 | resample-type: DOWN
11 | # add "classifier"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/collective_score/scorers/distance.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.collective_score.scorers.DistanceCollectiveAnswerScorer
2 |
3 | top-limit: [1, 3, 5, 10]
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/collective_score/scorers/edit-distance.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.collective_score.scorers.EditDistanceCollectiveAnswerScorer
2 |
3 | top-limit: [1, 3, 5, 10]
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/collective_score/scorers/original.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.collective_score.scorers.OriginalCollectiveAnswerScorer
2 |
3 | top-limit: [1, 3, 5, 10]
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/collective_score/scorers/shape-distance.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.collective_score.scorers.ShapeDistanceCollectiveAnswerScorer
2 |
3 | top-limit: [1, 3, 5, 10]
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/collective_score/scorers/type-coercion.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.collective_score.scorers.TypeCoercionCollectiveAnswerScorer
2 |
3 | top-limit: [1, 3, 5, 10]
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/generate/generate.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.generate.CavGenerationManager
2 |
3 | generators: |
4 | - inherit: baseqa.answer.generate.generators.choice
5 | - inherit: baseqa.answer.generate.generators.quantity
6 | - inherit: baseqa.answer.generate.generators.concept
7 | - inherit: baseqa.answer.generate.generators.cav-covering-concept
8 | # - inherit: baseqa.answer.generate.generators.covering-phrase
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/generate/generators/cav-covering-concept.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.generate.generators.CavCoveringConceptCavGenerator
2 |
3 | filter-question-concepts: false
4 | filter-question-tokens: false
5 | stoplist: null
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/generate/generators/choice.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.generate.generators.ChoiceCavGenerator
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/generate/generators/concept.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.generate.generators.ConceptCavGenerator
2 |
3 | filter-question-concepts: true
4 | filter-question-tokens: true
5 | stoplist: /dictionaries/stoplist-top5k.txt
6 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/generate/generators/covering-phrase.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.generate.generators.CoveringPhraseCavGenerator
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/generate/generators/quantity.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.generate.generators.QuantityCavGenerator
2 |
3 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/modify/modifiers/list-50.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.modify.modifiers.ListAnswerPruner
2 |
3 | threshold: 0.50
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/modify/modifiers/list-70.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.modify.modifiers.ListAnswerPruner
2 |
3 | threshold: 0.70
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/modify/modifiers/merge.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.modify.modifiers.CavMerger
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/modify/modifiers/ratio-72.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.modify.modifiers.ListAnswerPruner
2 |
3 | ratio: 0.72
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/modify/modifiers/ratio-88.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.modify.modifiers.ListAnswerPruner
2 |
3 | ratio: 0.88
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/modify/modify.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.modify.CavModificationManager
2 |
3 | handlers: |
4 | - inherit: baseqa.answer.modify.modifiers.merge
5 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/modify/pruner.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.modify.AnswerModificationManager
2 |
3 | handlers: |
4 | - inherit: baseqa.answer.modify.modifiers.list-70
5 | # - inherit: baseqa.answer.modify.modifiers.ratio-72
6 | # - inherit: baseqa.answer.modify.modifiers.ratio-88
7 | # - inherit: baseqa.answer.modify.modifiers.list-50
8 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/candidate-provider.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.AnswerCandidateProvider
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/classifier-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.learning_base.classifier-predict
2 |
3 | candidate-provider: 'inherit: baseqa.answer.score.candidate-provider'
4 | scorers: |
5 | - inherit: baseqa.answer.score.scorers.type-coercion
6 | - inherit: baseqa.answer.score.scorers.cao-count
7 | - inherit: baseqa.answer.score.scorers.name-count
8 | - inherit: baseqa.answer.score.scorers.avg-covered-token-count
9 | - inherit: baseqa.answer.score.scorers.stopword-count
10 | - inherit: baseqa.answer.score.scorers.token-overlap-count
11 | - inherit: baseqa.answer.score.scorers.concept-overlap-count
12 | - inherit: baseqa.answer.score.scorers.token-proximity
13 | - inherit: baseqa.answer.score.scorers.concept-proximity
14 | - inherit: baseqa.answer.score.scorers.focus-overlap-count
15 | - inherit: baseqa.answer.score.scorers.parsehead-proximity
16 | - inherit: baseqa.answer.score.scorers.focus
17 | - inherit: baseqa.answer.score.scorers.concept-type
18 | - inherit: baseqa.answer.score.scorers.answer-type
19 | # - inherit: baseqa.answer.score.scorers.parse
20 | # add "classifier", "feature-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/classifier-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.learning_base.classifier-train
2 |
3 | candidate-provider: 'inherit: baseqa.answer.score.candidate-provider'
4 | scorers: |
5 | - inherit: baseqa.answer.score.scorers.type-coercion
6 | - inherit: baseqa.answer.score.scorers.cao-count
7 | - inherit: baseqa.answer.score.scorers.name-count
8 | - inherit: baseqa.answer.score.scorers.avg-covered-token-count
9 | - inherit: baseqa.answer.score.scorers.stopword-count
10 | - inherit: baseqa.answer.score.scorers.token-overlap-count
11 | - inherit: baseqa.answer.score.scorers.concept-overlap-count
12 | - inherit: baseqa.answer.score.scorers.token-proximity
13 | - inherit: baseqa.answer.score.scorers.concept-proximity
14 | - inherit: baseqa.answer.score.scorers.focus-overlap-count
15 | - inherit: baseqa.answer.score.scorers.parsehead-proximity
16 | - inherit: baseqa.answer.score.scorers.focus
17 | - inherit: baseqa.answer.score.scorers.concept-type
18 | - inherit: baseqa.answer.score.scorers.answer-type
19 | # - inherit: baseqa.answer.score.scorers.parse
20 | resample-type: DOWN
21 | # add "classifier", "cv-predict-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/cv-load.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.learning_base.cv-load
2 |
3 | candidate-provider: 'inherit: baseqa.answer.score.candidate-provider'
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/answer-type.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.AnswerTypeAnswerScorer
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/avg-covered-token-count.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.AvgCoveredTokenCountAnswerScorer
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/cao-count.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.CaoCountAnswerScorer
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/concept-overlap-count.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.ConceptOverlappingCountAnswerScorer
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/concept-proximity.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.ConceptProximityAnswerScorer
2 |
3 | stoplist: /dictionaries/stoplist-top5k.txt
4 | window-size: 10
5 | infinity: .inf
6 | smoothing: 5.0
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/concept-type.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.ConceptTypeAnswerScorer
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/focus-overlap-count.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.FocusOverlappingCountAnswerScorer
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/focus-proximity.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.FocusProximityAnswerScorer
2 |
3 | window-size: 10
4 | infinity: .inf
5 | smoothing: 5.0
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/focus.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.FocusAnswerScorer
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/name-count.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.NameCountAnswerScorer
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/parse.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.ParseAnswerScorer
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/parsehead-proximity.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.ParseHeadProximityAnswerScorer
2 |
3 | stoplist: /dictionaries/stoplist-top5k.txt
4 | window-size: 10
5 | infinity: .inf
6 | smoothing: 5.0
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/stopword-count.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.StopwordCountAnswerScorer
2 |
3 | stoplist: /dictionaries/stoplist-top5k.txt
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/token-overlap-count.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.TokenOverlappingCountAnswerScorer
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/token-proximity.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.TokenProximityAnswerScorer
2 |
3 | stoplist: /dictionaries/stoplist-top5k.txt
4 | window-size: 10
5 | infinity: .inf
6 | smoothing: 5.0
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/scorers/type-coercion.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.scorers.TypeCoercionAnswerScorer
2 |
3 | at-limits: [1, 3]
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/score/simple.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.score.SimpleAnswerScorer
2 |
3 | type-coer-smoothing: 5.0
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/yesno/all-yes.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.yesno.AllYesYesNoAnswerPredictor
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/yesno/predict.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.yesno.YesNoAnswerPredictor
2 |
3 | scorers: |
4 | - inherit: baseqa.answer.yesno.scorers.concept-overlap
5 | - inherit: baseqa.answer.yesno.scorers.token-overlap
6 | - inherit: baseqa.answer.yesno.scorers.expected-answer-overlap
7 | - inherit: baseqa.answer.yesno.scorers.sentiment
8 | - inherit: baseqa.answer.yesno.scorers.negation
9 | # requires "classifier"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/yesno/scorers/alternate-answer.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.yesno.scorers.AlternateAnswerYesNoScorer
2 |
3 | # requires "pipeline"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/yesno/scorers/concept-overlap.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.yesno.scorers.ConceptOverlapYesNoScorer
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/yesno/scorers/expected-answer-overlap.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.yesno.scorers.ExpectedAnswerOverlapYesNoScorer
2 |
3 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/yesno/scorers/negation.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.yesno.scorers.NegationYesNoScorer
2 |
3 | negation-cues-path: /dictionaries/negation-cues.txt
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/yesno/scorers/sentiment.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.yesno.scorers.SentimentYesNoScorer
2 |
3 | positive-wordlist-path: /dictionaries/positive-words.txt
4 | negative-wordlist-path: /dictionaries/negative-words.txt
5 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/yesno/scorers/token-overlap.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.yesno.scorers.TokenOverlapYesNoScorer
2 |
3 | stoplist-path: /dictionaries/stoplist.txt
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer/yesno/train.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer.yesno.YesNoAnswerTrainer
2 |
3 | scorers: |
4 | - inherit: baseqa.answer.yesno.scorers.concept-overlap
5 | - inherit: baseqa.answer.yesno.scorers.token-overlap
6 | - inherit: baseqa.answer.yesno.scorers.expected-answer-overlap
7 | - inherit: baseqa.answer.yesno.scorers.sentiment
8 | - inherit: baseqa.answer.yesno.scorers.negation
9 | resample-type: DOWN
10 | # requires "classifier", "cv-predict-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer_type/feature-constructor.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.feature-constructor
2 |
3 | quantity-question-words-path: /dictionaries/quantity-question-words.txt
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer_type/gslabel-concept-search.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer_type.ConceptSearchGSAnswerTypeLabeler
2 |
3 | quantity-question-words-path: /dictionaries/quantity-question-words.txt
4 |
5 | # add values for parameters "concept-search-provider" and "at-gslabel-file"
6 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer_type/gslabel-concept.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer_type.ConceptGSAnswerTypeLabeler
2 |
3 | quantity-question-words-path: /dictionaries/quantity-question-words.txt
4 |
5 | # add values for parameters "concept-provider" and "at-gslabel-file"
6 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer_type/load-cv.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer_type.AnswerTypeCVPredictLoader
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer_type/predict.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer_type.AnswerTypeClassifierPredictor
2 |
3 | feature-constructor: 'inherit: baseqa.answer_type.feature-constructor'
4 | # requires parameter "model"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/answer_type/train.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.answer_type.AnswerTypeClassifierTrainer
2 |
3 | feature-constructor: 'inherit: baseqa.answer_type.feature-constructor'
4 | cv-predict-limit: 5
5 | # requires parameter "classifier", "at-gslabel-files", "null-type", "type-ratio-threshold"
6 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/cas-serialize.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.CasSerializer
2 |
3 | dir: result/
4 | typesystem: baseqa.type.OAQATypes
--------------------------------------------------------------------------------
/src/main/resources/baseqa/concept/rerank/candidate-provider.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.concept.rerank.ConceptSearchResultCandidateProvider
--------------------------------------------------------------------------------
/src/main/resources/baseqa/concept/rerank/classifier-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.learning_base.classifier-predict
2 |
3 | candidate-provider: 'inherit: baseqa.concept.rerank.candidate-provider'
4 | scorers: |
5 | - inherit: baseqa.concept.rerank.scorers.original-score
6 | - inherit: baseqa.concept.rerank.scorers.name-match
7 | # add "classifier", "feature-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/concept/rerank/classifier-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.learning_base.classifier-train
2 |
3 | candidate-provider: 'inherit: baseqa.concept.rerank.candidate-provider'
4 | scorers: |
5 | - inherit: baseqa.concept.rerank.scorers.original-score
6 | - inherit: baseqa.concept.rerank.scorers.name-match
7 | resample-type: DOWN
8 | # add "classifier", "cv-predict-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/concept/rerank/lucene.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.concept.rerank.LuceneInMemoryConceptReranker
2 |
3 | query-string-contructor: edu.cmu.lti.oaqa.baseqa.providers.query.LuceneQueryStringConstructor
4 | query-analyzer: org.apache.lucene.analysis.standard.StandardAnalyzer
5 | hits: 100
6 | limit: 10
7 | rerank-weight: 1.0
--------------------------------------------------------------------------------
/src/main/resources/baseqa/concept/rerank/name-match.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.concept.rerank.MatchingNameConceptReranker
2 |
3 | bonus: 100.0
--------------------------------------------------------------------------------
/src/main/resources/baseqa/concept/rerank/score-sum.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.concept.rerank.ScoreSummationConceptReranker
2 |
3 | limit: 5
--------------------------------------------------------------------------------
/src/main/resources/baseqa/concept/rerank/scorers/lucene.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.concept.rerank.scorers.LuceneConceptScorer
2 |
3 | stoplist-path: /dictionaries/stoplist.txt
4 | hits: 100
5 |
6 | # requires "fields", "index", "id-field", "source-field", "uri-prefix"
7 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/concept/rerank/scorers/name-match.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.concept.rerank.scorers.MatchingNameConceptScorer
--------------------------------------------------------------------------------
/src/main/resources/baseqa/concept/rerank/scorers/original-score.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.concept.rerank.scorers.OriginalScoreConceptScorer
--------------------------------------------------------------------------------
/src/main/resources/baseqa/concept/rerank/searchid-weight.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.concept.rerank.WeightingSearchIdConceptReranker
2 |
3 | # requires "weights-path" parameter
--------------------------------------------------------------------------------
/src/main/resources/baseqa/concept/retrieval/lucene.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.concept.retrieval.LuceneConceptRetrievalExecutor
2 |
3 | query-string-constructor: edu.cmu.lti.oaqa.baseqa.providers.query.LuceneQueryStringConstructor
4 | query-analyzer: org.apache.lucene.analysis.standard.StandardAnalyzer
5 | hits: 100
6 | # requires "fields", "index", "id-field", "name-field", "source-field", "uri-prefix"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/document/rerank/candidate-provider.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.document.rerank.DocumentCandidateProvider
--------------------------------------------------------------------------------
/src/main/resources/baseqa/document/rerank/classifier-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.learning_base.classifier-predict
2 |
3 | candidate-provider: 'inherit: baseqa.document.rerank.candidate-provider'
4 | scorers: |
5 | - inherit: baseqa.document.rerank.scorers.original-score
6 | # add "classifier", "feature-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/document/rerank/classifier-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.learning_base.classifier-train
2 |
3 | candidate-provider: 'inherit: baseqa.document.rerank.candidate-provider'
4 | scorers: |
5 | - inherit: baseqa.document.rerank.scorers.original-score
6 | resample-type: DOWN
7 | # add "classifier", "cv-predict-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/document/rerank/cv-load.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.learning_base.cv-load
2 |
3 | candidate-provider: 'inherit: baseqa.document.rerank.candidate-provider'
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/document/rerank/logreg.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.document.rerank.LogRegDocumentReranker
2 |
3 | query-string-contructor: edu.cmu.lti.oaqa.baseqa.providers.query.LuceneQueryStringConstructor
4 | query-analyzer: org.apache.lucene.analysis.standard.StandardAnalyzer
5 | hits: 100
6 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/document/rerank/scorers/lucene.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.document.rerank.scorers.LuceneDocumentScorer
2 |
3 | hits: 100
4 | stoplist-path: /dictionaries/stoplist.txt
5 | # requires "fields", "index", "id-field", "uri-prefix"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/document/rerank/scorers/original-score.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.document.rerank.scorers.OriginalScoreDocumentScorer
--------------------------------------------------------------------------------
/src/main/resources/baseqa/document/retrieval/lucene.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.document.retrieval.LuceneDocumentRetrievalExecutor
2 |
3 | query-string-constructor: edu.cmu.lti.oaqa.baseqa.providers.query.LuceneQueryStringConstructor
4 | query-analyzer: org.apache.lucene.analysis.standard.StandardAnalyzer
5 | hits: 100
6 | # requires "fields", "index", "id-field", "title-field", "text-field", "uri-prefix"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/evidence/concept/frequent-phrase.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.concept.PassageConceptRecognizer
2 |
3 | concept-provider: 'inherit: baseqa.providers.kb.frequent-phrase'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/baseqa/evidence/concept/merge.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.concept.ConceptMerger
2 |
3 | include-default-view: true
4 | view-name-prefix: ptv
5 | use-name: true
--------------------------------------------------------------------------------
/src/main/resources/baseqa/evidence/concept/opennlp-np.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.concept.PassageConceptRecognizer
2 |
3 | concept-provider: 'inherit: baseqa.providers.kb.opennlp-np'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/baseqa/evidence/concept/opennlp-npppnp.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.concept.PassageConceptRecognizer
2 |
3 | concept-provider: 'inherit: baseqa.providers.kb.opennlp-npppnp'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/baseqa/evidence/concept/search.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.concept.ConceptSearcher
2 |
3 | # add "concept-search-provider", "synonym-expansion-provider"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/evidence/parse/clearnlp-general.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.parse.PassageParser
2 |
3 | parser-provider: 'inherit: baseqa.providers.parser.clearnlp-general'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/baseqa/evidence/parse/lingpipe-indoeuro-brown.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.parse.PassageParser
2 |
3 | parser-provider: 'inherit: baseqa.providers.parser.lingpipe-indoeuro-brown'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/baseqa/evidence/parse/lingpipe-regex-brown.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.parse.PassageParser
2 |
3 | parser-provider: 'inherit: baseqa.providers.parser.lingpipe-regex-brown'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/baseqa/evidence/passage-to-view.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.PassageToViewCopier
2 |
3 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/baseqa/learning_base/classifier-predict.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.learning_base.ClassifierPredictor
2 |
3 | # requires "candidate-provider", "scorers", "classifier", "feature-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/learning_base/classifier-train.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.learning_base.ClassifierTrainer
2 |
3 | # requires "candidate-provider", "scorers", "classifier", "cv-predict-file", "at-least-one-correct", "resample-type"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/learning_base/cv-load.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.learning_base.CVPredictLoader
2 |
3 | # requires "candidate-provider", "cv-predict-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/passage/rerank/candidate-provider.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.passage.rerank.PassageCandidateProvider
--------------------------------------------------------------------------------
/src/main/resources/baseqa/passage/rerank/classifier-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.learning_base.classifier-predict
2 |
3 | candidate-provider: 'inherit: baseqa.passage.rerank.candidate-provider'
4 | scorers: |
5 | - inherit: baseqa.passage.rerank.scorers.original-score
6 | - inherit: baseqa.passage.rerank.scorers.meta-info
7 | - inherit: baseqa.passage.rerank.scorers.lucene
8 | # add "classifier", "feature-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/passage/rerank/classifier-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.learning_base.classifier-train
2 |
3 | candidate-provider: 'inherit: baseqa.passage.rerank.candidate-provider'
4 | scorers: |
5 | - inherit: baseqa.passage.rerank.scorers.original-score
6 | - inherit: baseqa.passage.rerank.scorers.meta-info
7 | - inherit: baseqa.passage.rerank.scorers.lucene
8 | resample-type: DOWN
9 | # add "classifier", "cv-predict-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/passage/rerank/scorers/lucene.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.passage.rerank.scorers.LuceneInMemoryPassageScorer
2 |
3 | hits: 100
4 | stoplist-path: /dictionaries/stoplist.txt
--------------------------------------------------------------------------------
/src/main/resources/baseqa/passage/rerank/scorers/meta-info.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.passage.rerank.scorers.MetaInfoPassageScorer
--------------------------------------------------------------------------------
/src/main/resources/baseqa/passage/rerank/scorers/original-score.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.passage.rerank.scorers.OriginalScorePassageScorer
--------------------------------------------------------------------------------
/src/main/resources/baseqa/passage/retrieval/document-to-passage.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.passage.retrieval.DocumentToPassageConverter
2 |
3 | include-title-abstract: true
4 | include-sections: true
5 | only-for-documents-missing-sections: true
6 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/passage/retrieval/lucene-sentence-improve.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.passage.retrieval.ImprovedLuceneInMemorySentenceRetrievalExecutor
2 |
3 | query-string-constructor: edu.cmu.lti.oaqa.baseqa.providers.query.LuceneQueryStringConstructor
4 | query-analyzer: org.apache.lucene.analysis.standard.StandardAnalyzer
5 | hits: 200
6 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/passage/retrieval/lucene-sentence.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.passage.retrieval.LuceneInMemorySentenceRetrievalExecutor
2 |
3 | query-string-constructor: edu.cmu.lti.oaqa.baseqa.providers.query.LuceneQueryStringConstructor
4 | query-analyzer: org.apache.lucene.analysis.standard.StandardAnalyzer
5 | hits: 200
6 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/preprocess/passage-concept-cache.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.preprocess.PassageConceptCache
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/preprocess/question-concept-cache.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.preprocess.QuestionConceptCache
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/kb/frequent-phrase.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.kb.FrequentPhraseConceptProvider
2 |
3 | max-length: 6
4 | min-freq: 2
5 | min-freq-ratio: 0.0
6 | type: AVERAGE
7 | score-threshold: 1.0
8 | score-ratio-threshold: 0.8
9 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/kb/opennlp-np.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.kb.opennlp
2 |
3 | type: NP
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/kb/opennlp-npppnp.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.kb.opennlp
2 |
3 | type: NP,PP,NP
4 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/kb/opennlp.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.kb.OpenNlpChunkerConceptProvider
2 |
3 | chunker-model: /en-chunker.bin
4 | min-length: 2
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/ml/classifiers/feature-constructor.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.ml.classifiers.FeatureConstructorProviderImpl
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/ml/classifiers/liblinear.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.ml.classifiers.LibLinearProvider
2 |
3 | # add parameters "feat-index-file", "label-index-file", "model-file", "balance-weight",
4 | # "solver-type"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/ml/classifiers/libsvm.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.ml.classifiers.LibSvmProvider
2 |
3 | # add parameters "feat-index-file", "label-index-file", "model-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/ml/classifiers/meka-pcc.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.meka
2 |
3 | classifier-name: meka.classifiers.multilabel.PCC
4 | options: [-W, weka.classifiers.meta.LogitBoost]
5 |
6 | # add parameters "model-file", "dataset-schema-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/ml/classifiers/meka.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.ml.classifiers.MekaProvider
2 |
3 | # add parameters "model-file", "classifier-name", "options", "dataset-schema-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/ml/classifiers/weka-cvr.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka
2 |
3 | classifier-name: weka.classifiers.meta.ClassificationViaRegression
4 | options: []
5 |
6 | #classifier-name: weka.classifiers.meta.AttributeSelectedClassifier
7 | #options:
8 | # - '-E'
9 | # - 'weka.attributeSelection.CorrelationAttributeEval'
10 | # - '-S'
11 | # - 'weka.attributeSelection.Ranker -T -1.7976931348623157E308 -N 100'
12 | # - '-W'
13 | # - 'weka.classifiers.functions.SimpleLogistic'
14 |
15 | #classifier-name: weka.classifiers.meta.ClassificationViaClustering
16 | #options: []
17 |
18 | # add parameters "model-file", "dataset-schema-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/ml/classifiers/weka-knn.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka
2 |
3 | classifier-name: weka.classifiers.lazy.IBk
4 | options: []
5 |
6 | # add parameters "model-file", "dataset-schema-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/ml/classifiers/weka-logistic.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka
2 |
3 | classifier-name: weka.classifiers.functions.SimpleLogistic
4 | options: []
5 |
6 | # add parameters "model-file", "dataset-schema-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/ml/classifiers/weka-reptree.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka
2 |
3 | classifier-name: weka.classifiers.trees.REPTree
4 | options: []
5 |
6 | # add parameters "model-file", "dataset-schema-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/ml/classifiers/weka.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.ml.classifiers.WekaProvider
2 |
3 | # add parameters "model-file", "classifier-name", "options", "dataset-schema-file"
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/parser/clearnlp-general.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.parser.ClearParserProvider
2 |
3 | pos-model: /general-en-pos.xz
4 | dep-model: /general-en-dep.xz
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/parser/lingpipe-indoeuro-brown.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.parser.LingPipeParserProvider
2 |
3 | token-factory: com.aliasi.tokenizer.IndoEuropeanTokenizerFactory
4 | token-factory-params: []
5 | pos-model: /pos-en-general-brown.HiddenMarkovModel
--------------------------------------------------------------------------------
/src/main/resources/baseqa/providers/parser/lingpipe-regex-brown.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.parser.LingPipeParserProvider
2 |
3 | token-factory: com.aliasi.tokenizer.RegExTokenizerFactory
4 | token-factory-params: ['([-''\d\p{L}]|\([-''\d\p{L}]+\))*[-''\d\p{L}]+|\S']
5 | pos-model: /pos-en-general-brown.HiddenMarkovModel
--------------------------------------------------------------------------------
/src/main/resources/baseqa/question/concept/opennlp-np.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.concept.QuestionConceptRecognizer
2 |
3 | concept-provider: 'inherit: baseqa.providers.kb.opennlp-np'
--------------------------------------------------------------------------------
/src/main/resources/baseqa/question/concept/opennlp-npppnp.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.concept.QuestionConceptRecognizer
2 |
3 | concept-provider: 'inherit: baseqa.providers.kb.opennlp-npppnp'
--------------------------------------------------------------------------------
/src/main/resources/baseqa/question/focus.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.focus.QuestionFocusExtractor
2 |
--------------------------------------------------------------------------------
/src/main/resources/baseqa/question/parse/clearnlp-general.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.parse.QuestionParser
2 |
3 | parser-provider: 'inherit: baseqa.providers.parser.clearnlp-general'
--------------------------------------------------------------------------------
/src/main/resources/baseqa/question/parse/lingpipe-indoeuro-brown.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.parse.QuestionParser
2 |
3 | parser-provider: 'inherit: baseqa.providers.parser.lingpipe-indoeuro-brown'
--------------------------------------------------------------------------------
/src/main/resources/baseqa/question/parse/lingpipe-regex-brown.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.parse.QuestionParser
2 |
3 | parser-provider: 'inherit: baseqa.providers.parser.lingpipe-regex-brown'
--------------------------------------------------------------------------------
/src/main/resources/baseqa/question/yesno/yesno-to-factoid.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.yesno.YesNoToFactoidQuestionConverter
--------------------------------------------------------------------------------
/src/main/resources/bioasq/collection/json/json-cas-consumer.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.collection.json.json-cas-consumer
2 |
3 | document-limit: 10
4 | snippet-limit: 10
5 | concept-limit: 10
6 | triple-limit: 10
7 | factoid-answer-limit: 5
8 | list-answer-limit: 100
9 |
10 | result-dir: result/
11 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/concept/rerank/classifier-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.concept.rerank.classifier-predict
2 |
3 | scorers: |
4 | - inherit: baseqa.concept.rerank.scorers.original-score
5 | - inherit: baseqa.concept.rerank.scorers.name-match
6 | - inherit: bioasq.concept.rerank.scorers.gopubmed
--------------------------------------------------------------------------------
/src/main/resources/bioasq/concept/rerank/classifier-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.concept.rerank.classifier-train
2 |
3 | scorers: |
4 | - inherit: baseqa.concept.rerank.scorers.original-score
5 | - inherit: baseqa.concept.rerank.scorers.name-match
6 | - inherit: bioasq.concept.rerank.scorers.gopubmed
--------------------------------------------------------------------------------
/src/main/resources/bioasq/concept/rerank/liblinear-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioasq.concept.rerank.classifier-predict
2 |
3 | classifier: 'inherit: bioasq.concept.rerank.liblinear'
4 | feature-file: result/concept-rerank-predict-liblinear.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioasq/concept/rerank/liblinear-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioasq.concept.rerank.classifier-train
2 |
3 | classifier: 'inherit: bioasq.concept.rerank.liblinear'
4 | cv-predict-file: src/main/resources/models/bioasq/concept/rerank/4b-dev-liblinear.cv
--------------------------------------------------------------------------------
/src/main/resources/bioasq/concept/rerank/liblinear.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.liblinear
2 |
3 | feat-index-file: src/main/resources/models/bioasq/concept/rerank/4b-dev-liblinear.findex
4 | label-index-file: src/main/resources/models/bioasq/concept/rerank/4b-dev-liblinear.lindex
5 | model-file: src/main/resources/models/bioasq/concept/rerank/4b-dev-liblinear.model
6 | solver-type: L1R_LR # L2R_L1LOSS_SVC_DUAL, L1R_LR, L2R_LR_DUAL, L2R_L2LOSS_SVC, L2R_L2LOSS_SVC_DUAL, L1R_L2LOSS_SVC, L2R_LR
7 | balance-weight: false
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/concept/rerank/ontology-logreg-params.txt:
--------------------------------------------------------------------------------
1 | DISEASE -4.9734 12.3044
2 | JOCHEM -9.6081 34.8384
3 | MESH -4.9894 12.9642
4 | UNIPROT -7.3101 13.8336
5 | GENE -3.0187 5.8931
--------------------------------------------------------------------------------
/src/main/resources/bioasq/concept/rerank/scorers/gopubmed.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioasq.concept.rerank.scorers.GoPubMedConceptRetrievalScorer
2 |
3 | pages: 1
4 | hits: 100
5 | conf: /properties/direct-gopubmed-concept.properties
6 | timeout: 2
7 | stoplist-path: /dictionaries/stoplist.txt
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/concept/rerank/searchid-weight.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.rerank.concept.searchid-weight
2 |
3 | weights-path: /bioasq/concept/rerank/ontology-logreg-params.txt
--------------------------------------------------------------------------------
/src/main/resources/bioasq/concept/retrieval/gopubmed-separate.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioasq.concept.retrieval.GoPubMedSeparateConceptRetrievalExecutor
2 |
3 | pages: 1
4 | hits: 10
5 | conf: /properties/direct-gopubmed-concept.properties # /properties/metal-gopubmed-concept.properties
6 | timeout: 1
7 | limit: 500
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/concept/retrieval/gopubmed.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioasq.concept.retrieval.GoPubMedConceptRetrievalExecutor
2 |
3 | pages: 1
4 | hits: 100
5 | conf: /properties/direct-gopubmed-concept.properties # /properties/metal-gopubmed-concept.properties
6 | timeout: 1 # 4
7 | limit: 500
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/document/retrieval/dixu-proprietary.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioasq.document.retrieval.DiXuProprietaryDocumentRetrievalClient
2 |
3 | host: 128.2.190.25
4 | port: 10008
5 | uri-prefix: http://www.ncbi.nlm.nih.gov/pubmed/
6 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/document/retrieval/gopubmed.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioasq.document.retrieval.GoPubMedDocumentRetrievalExecutor
2 |
3 | pages: 4
4 | hits: 100
5 | conf: /properties/direct-gopubmed-document.properties # /properties/metal-gopubmed-document.properties
--------------------------------------------------------------------------------
/src/main/resources/bioasq/eval/calculator/answer-eval-calculator.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioasq.eval.calculator.AnswerEvalCalculator
2 |
3 | # No parameter needed
4 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/gs/bioasq-qa-gs-decorator.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.collection.json.json-gs-decorator
2 |
3 | file: input/4b-dev.json.auto.fulltext
4 | #file: input/3b-dev.json.fixed
5 | #file: input/4b-dev.json.auto.abstract
--------------------------------------------------------------------------------
/src/main/resources/bioasq/passage/pmc-content.yaml.template:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioasq.passage.PmcContentSetter
2 |
3 | url-format: # ADD THE FORMATABLE URL HERE, E.G. http://localhost:8080/pmc/%s
4 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/passage/retrieval/document-to-passage.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.passage.retrieval.document-to-passage
2 |
3 | include-title-abstract: true
4 | include-sections: false
5 | only-for-documents-missing-sections: false
6 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/phase-a-test.yaml:
--------------------------------------------------------------------------------
1 | # execute
2 | # mvn exec:exec -Dconfig=bioasq.phase-a-test
3 | # to test the pipeline
4 |
5 | configuration:
6 | name: phase-a-test
7 | author: ziy
8 |
9 | persistence-provider:
10 | inherit: baseqa.persistence.local-sqlite-persistence-provider
11 |
12 | collection-reader:
13 | inherit: baseqa.collection.json.json-collection-reader
14 | dataset: BIOASQ-QA
15 | file:
16 | # - input/4b-1-a.json
17 | # - input/4b-2-a.json
18 | # - input/4b-3-a.json
19 | # - input/4b-4-a.json
20 | - input/4b-5-a.json
21 | persistence-provider: |
22 | inherit: baseqa.persistence.local-sqlite-persistence-provider
23 |
24 | pipeline:
25 | - inherit: ecd.phase
26 | name: question-parse
27 | options: |
28 | - inherit: bioqa.question.parse.clearnlp-bioinformatics
29 |
30 | - inherit: ecd.phase
31 | name: question-concept-metamap
32 | options: |
33 | - inherit: bioqa.question.concept.metamap-cached
34 |
35 | - inherit: ecd.phase
36 | name: question-concept-tmtool
37 | options: |
38 | - inherit: bioqa.question.concept.tmtool-cached
39 |
40 | - inherit: ecd.phase
41 | name: question-concept-lingpipe-genia
42 | options: |
43 | - inherit: bioqa.question.concept.lingpipe-genia
44 |
45 | - inherit: ecd.phase
46 | name: concept-search-uts
47 | options: |
48 | - inherit: bioqa.evidence.concept.search-uts-cached
49 |
50 | - inherit: ecd.phase
51 | name: concept-merge
52 | options: |
53 | - inherit: baseqa.evidence.concept.merge
54 |
55 | - inherit: ecd.phase
56 | name: abstract-query-primary
57 | options: |
58 | - inherit: baseqa.abstract_query.token-concept
59 |
60 | # concept
61 | - inherit: ecd.phase
62 | name: concept-retrieval
63 | options: |
64 | - inherit: bioqa.concept.retrieval.lucene-bioconcept
65 |
66 | - inherit: ecd.phase
67 | name: concept-rerank
68 | options: |
69 | - inherit: bioqa.concept.rerank.liblinear-predict
70 |
71 | # document
72 | - inherit: ecd.phase
73 | name: document-retrieval
74 | options: |
75 | - inherit: bioqa.document.retrieval.lucene-medline
76 |
77 | - inherit: ecd.phase
78 | name: document-rerank
79 | options: |
80 | - inherit: bioqa.document.rerank.liblinear-predict
81 |
82 | # snippet
83 | - inherit: ecd.phase
84 | name: passage-retrieval
85 | options: |
86 | - inherit: bioasq.passage.retrieval.document-to-passage
87 |
88 | - inherit: ecd.phase
89 | name: passage-rerank
90 | options: |
91 | - inherit: bioqa.passage.rerank.liblinear-predict
92 | - inherit: base.noop
93 |
94 | post-process:
95 | # submission
96 | - inherit: bioasq.collection.json.json-cas-consumer
97 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/phase-a-train-snippet.yaml:
--------------------------------------------------------------------------------
1 | # execute
2 | # mvn exec:exec -Dconfig=bioasq.phase-a-train-snippet
3 | # to test the pipeline
4 |
5 | configuration:
6 | name: phase-a-train-snippet
7 | author: ziy
8 |
9 | persistence-provider:
10 | inherit: baseqa.persistence.local-sqlite-persistence-provider
11 |
12 | collection-reader:
13 | inherit: baseqa.collection.json.json-collection-reader
14 | dataset: BIOASQ-QA
15 | file:
16 | - input/dryrun-a.json
17 | - input/1b-1-a.json
18 | - input/1b-2-a.json
19 | - input/1b-3-a.json
20 | - input/2b-1-a.json
21 | - input/2b-2-a.json
22 | - input/2b-3-a.json
23 | - input/2b-4-a.json
24 | - input/2b-5-a.json
25 | - input/3b-1-a.json
26 | - input/3b-2-a.json
27 | - input/3b-3-a.json
28 | - input/3b-4-a.json
29 | # - input/3b-5-a.json
30 | decorators: |
31 | - inherit: bioasq.gs.bioasq-qa-gs-decorator
32 | persistence-provider: |
33 | inherit: baseqa.persistence.local-sqlite-persistence-provider
34 |
35 | pipeline:
36 | - inherit: ecd.phase
37 | name: question-parse
38 | options: |
39 | - inherit: bioqa.question.parse.clearnlp-bioinformatics
40 |
41 | - inherit: ecd.phase
42 | name: question-concept-metamap
43 | options: |
44 | - inherit: bioqa.question.concept.metamap-cached
45 |
46 | - inherit: ecd.phase
47 | name: question-concept-tmtool
48 | options: |
49 | - inherit: bioqa.question.concept.tmtool-cached
50 |
51 | - inherit: ecd.phase
52 | name: question-concept-lingpipe-genia
53 | options: |
54 | - inherit: bioqa.question.concept.lingpipe-genia
55 |
56 | - inherit: ecd.phase
57 | name: concept-search-uts
58 | options: |
59 | - inherit: bioqa.evidence.concept.search-uts-cached
60 |
61 | - inherit: ecd.phase
62 | name: concept-merge
63 | options: |
64 | - inherit: baseqa.evidence.concept.merge
65 |
66 | - inherit: ecd.phase
67 | name: abstract-query-primary
68 | options: |
69 | - inherit: baseqa.abstract_query.token-concept
70 |
71 | # document
72 | - inherit: ecd.phase
73 | name: document-retrieval
74 | options: |
75 | - inherit: bioqa.document.retrieval.lucene-medline
76 |
77 | - inherit: ecd.phase
78 | name: document-rerank
79 | options: |
80 | - inherit: bioqa.document.rerank.cv-load-liblinear
81 |
82 | # snippet
83 | - inherit: ecd.phase
84 | name: passage-retrieval
85 | options: |
86 | - inherit: bioasq.passage.retrieval.document-to-passage
87 |
88 | - inherit: ecd.phase
89 | name: passage-rerank-train-liblinear
90 | options: |
91 | - inherit: bioqa.passage.rerank.liblinear-train
92 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/phase-b-test-yesno.yaml:
--------------------------------------------------------------------------------
1 | # execute
2 | # mvn exec:exec -Dconfig=bioasq.phase-b-test-yesno
3 | # to test the pipeline
4 |
5 | configuration:
6 | name: phase-b-test-yesno
7 | author: ziy
8 |
9 | persistence-provider:
10 | inherit: baseqa.persistence.local-sqlite-persistence-provider
11 |
12 | collection-reader:
13 | inherit: baseqa.collection.json.json-collection-reader
14 | dataset: BIOASQ-QA
15 | file:
16 | - input/4b-5-b.json
17 | # - input/4b-all-b-no.json
18 | type: [yesno]
19 | persistence-provider: |
20 | inherit: baseqa.persistence.local-sqlite-persistence-provider
21 |
22 | pipeline:
23 | - inherit: ecd.phase
24 | name: question-parse
25 | options: |
26 | - inherit: bioqa.question.parse.clearnlp-bioinformatics
27 |
28 | - inherit: ecd.phase
29 | name: question-concept-metamap
30 | options: |
31 | - inherit: bioqa.question.concept.metamap-cached
32 |
33 | - inherit: ecd.phase
34 | name: question-concept-tmtool
35 | options: |
36 | - inherit: bioqa.question.concept.tmtool-cached
37 |
38 | - inherit: ecd.phase
39 | name: question-concept-lingpipe-genia
40 | options: |
41 | - inherit: bioqa.question.concept.lingpipe-genia
42 |
43 | - inherit: ecd.phase
44 | name: passage-to-view
45 | options: |
46 | - inherit: baseqa.evidence.passage-to-view
47 |
48 | - inherit: ecd.phase
49 | name: evidence-parse
50 | options: |
51 | - inherit: bioqa.evidence.parse.clearnlp-bioinformatics
52 |
53 | - inherit: ecd.phase
54 | name: evidence-concept-metamap
55 | options: |
56 | - inherit: bioqa.evidence.concept.metamap-cached
57 |
58 | - inherit: ecd.phase
59 | name: evidence-concept-tmtool
60 | options: |
61 | - inherit: bioqa.evidence.concept.tmtool-cached
62 |
63 | - inherit: ecd.phase
64 | name: evidence-concept-lingpipe-genia
65 | options: |
66 | - inherit: bioqa.evidence.concept.lingpipe-genia
67 |
68 | - inherit: ecd.phase
69 | name: evidence-concept-frequent-phrase
70 | options: |
71 | - inherit: baseqa.evidence.concept.frequent-phrase
72 |
73 | - inherit: ecd.phase
74 | name: concept-search-uts
75 | options: |
76 | - inherit: bioqa.evidence.concept.search-uts-cached
77 |
78 | - inherit: ecd.phase
79 | name: concept-merge
80 | options: |
81 | - inherit: baseqa.evidence.concept.merge
82 |
83 | - inherit: ecd.phase
84 | name: answer-yesno
85 | options: |
86 | - inherit: bioqa.answer.yesno.weka-cvr-predict
87 | # - inherit: baseqa.answer.yesno.all-yes
88 |
89 | # - inherit: baseqa.cas-serialize
90 |
91 | post-process:
92 | # submission
93 | - inherit: bioasq.collection.json.json-cas-consumer
94 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/phase-b-train-answer-type.yaml:
--------------------------------------------------------------------------------
1 | # execute
2 | # mvn exec:exec -Dconfig=bioasq.phase-b-train-answer-type
3 | # to test the pipeline
4 |
5 | configuration:
6 | name: phase-b-train-answer-type
7 | author: ziy
8 |
9 | persistence-provider:
10 | inherit: baseqa.persistence.local-sqlite-persistence-provider
11 |
12 | collection-reader:
13 | inherit: baseqa.collection.json.json-collection-reader
14 | dataset: BIOASQ-QA
15 | file:
16 | - input/dryrun-a.json
17 | - input/1b-1-a.json
18 | - input/1b-2-a.json
19 | - input/1b-3-a.json
20 | - input/2b-1-a.json
21 | - input/2b-2-a.json
22 | - input/2b-3-a.json
23 | - input/2b-4-a.json
24 | - input/2b-5-a.json
25 | - input/3b-1-a.json
26 | - input/3b-2-a.json
27 | - input/3b-3-a.json
28 | - input/3b-4-a.json
29 | # - input/3b-5-a.json
30 | type: [factoid, list]
31 | decorators: |
32 | - inherit: bioasq.gs.bioasq-qa-gs-decorator
33 | persistence-provider: |
34 | inherit: baseqa.persistence.local-sqlite-persistence-provider
35 |
36 | pipeline:
37 | - inherit: ecd.phase
38 | name: question-parse
39 | options: |
40 | - inherit: bioqa.question.parse.clearnlp-bioinformatics
41 |
42 | - inherit: ecd.phase
43 | name: question-concept-metamap
44 | options: |
45 | - inherit: bioqa.question.concept.metamap-cached
46 |
47 | - inherit: ecd.phase
48 | name: question-concept-tmtool
49 | options: |
50 | - inherit: bioqa.question.concept.tmtool-cached
51 |
52 | - inherit: ecd.phase
53 | name: question-concept-lingpipe-genia
54 | options: |
55 | - inherit: bioqa.question.concept.lingpipe-genia
56 |
57 | - inherit: ecd.phase
58 | name: question-focus
59 | options: |
60 | - inherit: baseqa.question.focus
61 |
62 | - inherit: ecd.phase
63 | name: concept-search-uts
64 | options: |
65 | - inherit: bioqa.evidence.concept.search-uts-cached
66 |
67 | - inherit: ecd.phase
68 | name: concept-merge
69 | options: |
70 | - inherit: baseqa.evidence.concept.merge
71 |
72 | - inherit: ecd.phase
73 | name: answer-type
74 | options: |
75 | - inherit: bioqa.answer_type.liblinear-train
76 | - inherit: bioqa.answer_type.liblinear-null-train
77 | # - inherit: bioqa.answer_type.train-weka-reptree
78 | # - inherit: bioqa.answer_type.train-weka-reptree-null
79 | # - inherit: bioqa.answer_type.train-weka-knn
80 | # - inherit: bioqa.answer_type.train-weka-knn-null
81 | # - inherit: bioqa.answer_type.train-meka-pcc
82 | # - inherit: bioqa.answer_type.train-meka-pcc-null
83 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/preprocess-answer-type-gslabel.yaml:
--------------------------------------------------------------------------------
1 | # execute
2 | # mvn exec:exec -Dconfig=bioasq.preprocess-answer-type-gslabel
3 | # to test the pipeline
4 |
5 | configuration:
6 | name: preprocess-answer-type-gslabel
7 | author: ziy
8 |
9 | persistence-provider:
10 | inherit: baseqa.persistence.local-sqlite-persistence-provider
11 |
12 | collection-reader:
13 | inherit: baseqa.collection.json.json-collection-reader
14 | dataset: BIOASQ-QA
15 | file:
16 | - input/dryrun-b.json
17 | - input/1b-1-b.json
18 | - input/1b-2-b.json
19 | - input/1b-3-b.json
20 | - input/2b-1-b.json
21 | - input/2b-2-b.json
22 | - input/2b-3-b.json
23 | - input/2b-4-b.json
24 | - input/2b-5-b.json
25 | - input/3b-1-b.json
26 | - input/3b-2-b.json
27 | - input/3b-3-b.json
28 | - input/3b-4-b.json
29 | - input/3b-5-b.json
30 | type: [factoid, list]
31 | decorators: |
32 | - inherit: bioasq.gs.bioasq-qa-gs-decorator
33 | persistence-provider: |
34 | inherit: baseqa.persistence.local-sqlite-persistence-provider
35 |
36 | pipeline:
37 | - inherit: ecd.phase
38 | name: question-parse
39 | options: |
40 | - inherit: bioqa.question.parse.clearnlp-bioinformatics
41 |
42 | - inherit: ecd.phase
43 | name: answer-type-gslabel-uts
44 | options: |
45 | - inherit: bioqa.answer_type.gslabel-uts
46 |
47 | - inherit: ecd.phase
48 | name: answer-type-gslabel-tmtool
49 | option-timeout: 30
50 | options: |
51 | - inherit: bioqa.answer_type.gslabel-tmtool
--------------------------------------------------------------------------------
/src/main/resources/bioasq/preprocess-kb-cache.yaml:
--------------------------------------------------------------------------------
1 | # execute
2 | # mvn exec:exec -Dconfig=bioasq.preprocess-kb-cache
3 | # to test the pipeline
4 |
5 | configuration:
6 | name: preprocess-kb-cache
7 | author: ziy
8 |
9 | persistence-provider:
10 | inherit: baseqa.persistence.local-sqlite-persistence-provider
11 |
12 | collection-reader:
13 | inherit: baseqa.collection.json.json-collection-reader
14 | dataset: BIOASQ-QA
15 | file:
16 | # - input/one-question.json
17 | - input/dryrun-b.json
18 | - input/1b-1-b.json
19 | - input/1b-2-b.json
20 | - input/1b-3-b.json
21 | - input/2b-1-b.json
22 | - input/2b-2-b.json
23 | - input/2b-3-b.json
24 | - input/2b-4-b.json
25 | - input/2b-5-b.json
26 | - input/3b-1-b.json
27 | - input/3b-2-b.json
28 | - input/3b-3-b.json
29 | - input/3b-4-b.json
30 | - input/3b-5-b.json
31 | - input/4b-1-b.json
32 | - input/4b-2-b.json
33 | - input/4b-3-b.json
34 | - input/4b-4-b.json
35 | - input/4b-5-b.json
36 | #- input/4b-dev.json.auto.abstract
37 | decorators: |
38 | - inherit: bioasq.gs.bioasq-qa-gs-decorator
39 | persistence-provider: |
40 | inherit: baseqa.persistence.local-sqlite-persistence-provider
41 |
42 | pipeline:
43 | - inherit: ecd.phase
44 | name: question-metamap-cache
45 | option-timeout: 30
46 | options: |
47 | - inherit: bioqa.preprocess.question-kb-cache
48 |
49 | - inherit: ecd.phase
50 | name: passage-metamap-cache
51 | option-timeout: 30
52 | options: |
53 | - inherit: bioqa.preprocess.passage-kb-cache
54 |
--------------------------------------------------------------------------------
/src/main/resources/bioasq/triple/retrieval/gopubmed.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioasq.triple.retrieval.GoPubMedTripleRetrievalExecutor
2 |
3 | pages: 1
4 | hits: 10
5 | conf: /properties/direct-gopubmed-triple.properties # /properties/metal-gopubmed-triple.properties
6 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/abstract_query/token-concept.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.abstract_query.token-concept
2 |
3 | stoplist-path: /dictionaries/stoplist-gene-top5k.txt
4 | use-weight: false
5 | use-type: false
6 | #pos-tags-path: /dictionaries/pos-tags.txt
7 | #noun-tags-path: /dictionaries/noun-tags.txt
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/cao/predict-crf.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.cao.predict
2 |
3 | transducer: 'inherit: bioqa.answer.cao.transducer-crf'
4 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/cao/predict-hmm.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.cao.predict
2 |
3 | transducer: 'inherit: bioqa.answer.cao.transducer-hmm'
4 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/cao/train-crf.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.cao.train
2 |
3 | transducer: 'inherit: bioqa.answer.cao.transducer-crf'
4 | sample-o2o: 1.0
5 | window-size: 20
6 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/cao/train-hmm.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.cao.train
2 |
3 | transducer: 'inherit: bioqa.answer.cao.transducer-hmm'
4 | sample-o2o: 1.0
5 | window-size: 50
6 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/cao/transducer-crf.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.transducers.crf
2 |
3 | model-file: src/main/resources/models/bioqa/answer/cao/3b-dev-crf-model
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/cao/transducer-hmm.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.transducers.hmm
2 |
3 | model-file: src/main/resources/models/bioqa/answer/cao/3b-dev-hmm-model
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/collective_score/liblinear-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.collective_score.classifier-predict
2 |
3 | classifier: 'inherit: bioqa.answer.collective_score.liblinear'
4 | feature-file: result/answer-collective-score-predict-liblinear.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/collective_score/liblinear-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.collective_score.classifier-train
2 |
3 | classifier: 'inherit: bioqa.answer.collective_score.liblinear'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/collective_score/liblinear.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.liblinear
2 |
3 | feat-index-file: src/main/resources/models/bioqa/answer/collective_score/4b-dev-liblinear.findex
4 | label-index-file: src/main/resources/models/bioqa/answer/collective_score/4b-dev-liblinear.lindex
5 | model-file: src/main/resources/models/bioqa/answer/collective_score/4b-dev-liblinear.model
6 | solver-type: L1R_LR # L2R_LR, L2R_L1LOSS_SVC_DUAL, L1R_LR, L2R_LR_DUAL, L2R_L2LOSS_SVC, L2R_L2LOSS_SVC_DUAL, L1R_L2LOSS_SVC
7 | balance-weight: false
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/collective_score/weka-cvr-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.collective_score.classifier-predict
2 |
3 | classifier: 'inherit: bioqa.answer.collective_score.weka-cvr'
4 | feature-file: result/answer-collective-score-predict-weka-cvr.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/collective_score/weka-cvr-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.collective_score.classifier-train
2 |
3 | classifier: 'inherit: bioqa.answer.collective_score.weka-cvr'
4 | resample-type: DOWN
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/collective_score/weka-cvr.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka-cvr
2 |
3 | model-file: src/main/resources/models/bioqa/answer/collective_score/4b-dev-weka-cvr.model
4 | dataset-schema-file: src/main/resources/models/bioqa/answer/collective_score/4b-dev-weka-cvr.dataset-schema
5 | balance-weight: true
6 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/generate/generate.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.generate.generate
2 |
3 | generators: |
4 | - inherit: baseqa.answer.generate.generators.choice
5 | - inherit: baseqa.answer.generate.generators.quantity
6 | - inherit: bioqa.answer.generate.generators.concept
7 | - inherit: baseqa.answer.generate.generators.cav-covering-concept
8 | # - inherit: baseqa.answer.generate.generators.covering-phrase
9 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/generate/generators/concept.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.generate.generators.concept
2 |
3 | filter-question-concepts: false
4 | filter-question-tokens: false
5 | stoplist: null # /dictionaries/stoplist-gene-top5k.txt # stoplist-gene-modified.txt, stoplist-top5k.txt
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/classifier-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.score.classifier-predict
2 |
3 | scorers: |
4 | - inherit: baseqa.answer.score.scorers.type-coercion
5 | - inherit: baseqa.answer.score.scorers.cao-count
6 | - inherit: baseqa.answer.score.scorers.name-count
7 | - inherit: baseqa.answer.score.scorers.avg-covered-token-count
8 | - inherit: bioqa.answer.score.scorers.stopword-count
9 | - inherit: baseqa.answer.score.scorers.token-overlap-count
10 | - inherit: baseqa.answer.score.scorers.concept-overlap-count
11 | - inherit: bioqa.answer.score.scorers.token-proximity
12 | - inherit: bioqa.answer.score.scorers.concept-proximity
13 | - inherit: baseqa.answer.score.scorers.focus-overlap-count
14 | - inherit: bioqa.answer.score.scorers.parsehead-proximity
15 | - inherit: baseqa.answer.score.scorers.answer-type
16 | - inherit: baseqa.answer.score.scorers.focus
17 | - inherit: baseqa.answer.score.scorers.concept-type
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/classifier-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.score.classifier-train
2 |
3 | scorers: |
4 | - inherit: baseqa.answer.score.scorers.type-coercion
5 | - inherit: baseqa.answer.score.scorers.cao-count
6 | - inherit: baseqa.answer.score.scorers.name-count
7 | - inherit: baseqa.answer.score.scorers.avg-covered-token-count
8 | - inherit: bioqa.answer.score.scorers.stopword-count
9 | - inherit: baseqa.answer.score.scorers.token-overlap-count
10 | - inherit: baseqa.answer.score.scorers.concept-overlap-count
11 | - inherit: bioqa.answer.score.scorers.token-proximity
12 | - inherit: bioqa.answer.score.scorers.concept-proximity
13 | - inherit: baseqa.answer.score.scorers.focus-overlap-count
14 | - inherit: bioqa.answer.score.scorers.parsehead-proximity
15 | - inherit: baseqa.answer.score.scorers.answer-type
16 | - inherit: baseqa.answer.score.scorers.focus
17 | - inherit: baseqa.answer.score.scorers.concept-type
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/cv-load-liblinear.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.score.cv-load
2 |
3 | cv-predict-file: /models/bioqa/answer/score/4b-dev-liblinear.cv
4 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/liblinear-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.answer.score.classifier-predict
2 |
3 | classifier: 'inherit: bioqa.answer.score.liblinear'
4 | feature-file: result/answer-score-predict-liblinear.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/liblinear-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.answer.score.classifier-train
2 |
3 | classifier: 'inherit: bioqa.answer.score.liblinear'
4 | cv-predict-file: src/main/resources/models/bioqa/answer/score/4b-dev-liblinear.cv
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/liblinear.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.liblinear
2 |
3 | feat-index-file: src/main/resources/models/bioqa/answer/score/4b-dev-liblinear.findex
4 | label-index-file: src/main/resources/models/bioqa/answer/score/4b-dev-liblinear.lindex
5 | model-file: src/main/resources/models/bioqa/answer/score/4b-dev-liblinear.model
6 | solver-type: L1R_LR # L2R_LR, L2R_L1LOSS_SVC_DUAL, L1R_LR, L2R_LR_DUAL, L2R_L2LOSS_SVC, L2R_L2LOSS_SVC_DUAL, L1R_L2LOSS_SVC
7 | balance-weight: true
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/scorers/concept-proximity.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.score.scorers.concept-proximity
2 |
3 | stoplist: /dictionaries/stoplist-gene-top5k.txt
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/scorers/parsehead-proximity.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.score.scorers.parsehead-proximity
2 |
3 | stoplist: /dictionaries/stoplist-gene-top5k.txt
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/scorers/stopword-count.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.score.scorers.stopword-count
2 |
3 | stoplist: /dictionaries/stoplist-gene-top5k.txt
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/scorers/token-proximity.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.score.scorers.token-proximity
2 |
3 | stoplist: /dictionaries/stoplist-gene-top5k.txt
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/weka-cvr-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.answer.score.classifier-predict
2 |
3 | classifier: 'inherit: bioqa.answer.score.weka-cvr'
4 | feature-file: result/answer-score-predict-weka-cvr.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/weka-cvr-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.answer.score.classifier-train
2 |
3 | classifier: 'inherit: bioqa.answer.score.weka-cvr'
4 | cv-predict-file: src/main/resources/models/bioqa/answer/score/4b-dev-weka-cvr.cv
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/score/weka-cvr.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka-cvr
2 |
3 | model-file: src/main/resources/models/bioqa/answer/score/4b-dev-weka-cvr.model
4 | dataset-schema-file: src/main/resources/models/bioqa/answer/score/4b-dev-weka-cvr.dataset-schema
5 | balance-weight: true
6 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/liblinear-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.answer.yesno.predict
2 |
3 | classifier: 'inherit: bioqa.answer.yesno.liblinear'
4 | feature-file: result/answer-yesno-predict-liblinear.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/liblinear-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.answer.yesno.train
2 |
3 | classifier: 'inherit: bioqa.answer.yesno.liblinear'
4 | cv-predict-file: src/main/resources/models/bioqa/answer/yesno/4b-dev-liblinear.cv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/liblinear.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.liblinear
2 |
3 | feat-index-file: src/main/resources/models/bioqa/answer/yesno/4b-dev-liblinear.findex
4 | label-index-file: src/main/resources/models/bioqa/answer/yesno/4b-dev-liblinear.lindex
5 | model-file: src/main/resources/models/bioqa/answer/yesno/4b-dev-liblinear.model
6 | solver-type: L1R_LR # L2R_LR, L2R_L1LOSS_SVC_DUAL, L1R_LR, L2R_LR_DUAL, L2R_L2LOSS_SVC, L2R_L2LOSS_SVC_DUAL, L1R_L2LOSS_SVC
7 | balance-weight: true
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.yesno.predict
2 |
3 | scorers: |
4 | - inherit: baseqa.answer.yesno.scorers.concept-overlap
5 | - inherit: bioqa.answer.yesno.scorers.token-overlap
6 | - inherit: baseqa.answer.yesno.scorers.expected-answer-overlap
7 | - inherit: baseqa.answer.yesno.scorers.sentiment
8 | - inherit: baseqa.answer.yesno.scorers.negation
9 | - inherit: bioqa.answer.yesno.scorers.alternate-answer
10 | # requires "classifier"
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/scorers/alternate-answer.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.yesno.scorers.alternate-answer
2 |
3 | factoid-pipeline: |
4 | - inherit: baseqa.question.yesno.yesno-to-factoid
5 | - inherit: bioqa.answer.generate.generate
6 | - inherit: baseqa.answer.modify.modify
7 | - inherit: bioqa.answer.score.liblinear-predict
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/scorers/token-overlap.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.yesno.scorers.token-overlap
2 |
3 | stoplist-path: /dictionaries/stoplist-gene-top5k.txt
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer.yesno.train
2 |
3 | scorers: |
4 | - inherit: baseqa.answer.yesno.scorers.concept-overlap
5 | - inherit: bioqa.answer.yesno.scorers.token-overlap
6 | - inherit: baseqa.answer.yesno.scorers.expected-answer-overlap
7 | - inherit: baseqa.answer.yesno.scorers.sentiment
8 | - inherit: baseqa.answer.yesno.scorers.negation
9 | - inherit: bioqa.answer.yesno.scorers.alternate-answer
10 | # requires "classifier", "cv-predict-file"
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/weka-cvr-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.answer.yesno.predict
2 |
3 | classifier: 'inherit: bioqa.answer.yesno.weka-cvr'
4 | feature-file: result/answer-yesno-predict-weka-cvr.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/weka-cvr-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.answer.yesno.train
2 |
3 | classifier: 'inherit: bioqa.answer.yesno.weka-cvr'
4 | cv-predict-file: src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-cvr.cv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/weka-cvr.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka-cvr
2 |
3 | model-file: src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-cvr.model
4 | dataset-schema-file: src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-cvr.dataset-schema
5 | dataset-export: src/main/resources/models/bioqa/answer/yesno/4b-dev-weka.arff
6 | balance-weight: true
7 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/weka-logistic-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.answer.yesno.predict
2 |
3 | classifier: 'inherit: bioqa.answer.yesno.weka-logistic'
4 | feature-file: result/answer-yesno-predict-weka-logistic.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/weka-logistic-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.answer.yesno.train
2 |
3 | classifier: 'inherit: bioqa.answer.yesno.weka-logistic'
4 | cv-predict-file: src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-logistic.cv
5 | resample-type: UP
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer/yesno/weka-logistic.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka-logistic
2 |
3 | model-file: src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-logistic.model
4 | dataset-schema-file: src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-logistic.dataset-schema
5 | dataset-export: src/main/resources/models/bioqa/answer/yesno/4b-dev-weka.arff
6 | balance-weight: true
7 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/gslabel-tmtool.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.gslabel-concept
2 |
3 | at-gslabel-file: src/main/resources/models/bioqa/answer_type/4b-dev-gslabel-tmtool.json
4 | concept-provider: 'inherit: bioqa.providers.kb.tmtool-cached'
5 | batch-size: 100
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/gslabel-uts.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.gslabel-concept-search
2 |
3 | at-gslabel-file: src/main/resources/models/bioqa/answer_type/4b-dev-gslabel-uts.json
4 | concept-search-provider: 'inherit: bioqa.providers.kb.concept-search-uts-cached'
5 | batch-size: 1
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/liblinear-null-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.predict
2 |
3 | classifier: 'inherit: bioqa.answer_type.liblinear-null'
4 | predict-file: result/answer-type-predict-liblinear-null.tsv
5 | limit: 5
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/liblinear-null-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.train
2 |
3 | classifier: 'inherit: bioqa.answer_type.liblinear-null'
4 | cv-predict-file: src/main/resources/models/bioqa/answer_type/4b-dev-liblinear-null.cv
5 | at-gslabel-files:
6 | - /models/bioqa/answer_type/4b-dev-gslabel-uts.json
7 | - /models/bioqa/answer_type/4b-dev-gslabel-tmtool.json
8 | null-type: true
9 | type-ratio-threshold: 0.5
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/liblinear-null.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.liblinear
2 |
3 | feat-index-file: src/main/resources/models/bioqa/answer_type/4b-dev-liblinear-null.findex
4 | label-index-file: src/main/resources/models/bioqa/answer_type/4b-dev-liblinear-null.lindex
5 | model-file: src/main/resources/models/bioqa/answer_type/4b-dev-liblinear-null.model
6 | solver-type: L1R_LR # L2R_L1LOSS_SVC_DUAL, L1R_LR, L2R_LR_DUAL, L2R_L2LOSS_SVC, L2R_L2LOSS_SVC_DUAL, L1R_L2LOSS_SVC, L2R_LR
7 | balance-weight: false
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/liblinear-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.predict
2 |
3 | classifier: 'inherit: bioqa.answer_type.liblinear'
4 | predict-file: result/answer-type-predict-liblinear.tsv
5 |
6 | cross-opts:
7 | limit: [5]
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/liblinear-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.train
2 |
3 | classifier: 'inherit: bioqa.answer_type.liblinear'
4 | cv-predict-file: src/main/resources/models/bioqa/answer_type/4b-dev-liblinear.cv
5 | at-gslabel-files:
6 | - /models/bioqa/answer_type/4b-dev-gslabel-uts.json
7 | - /models/bioqa/answer_type/4b-dev-gslabel-tmtool.json
8 | null-type: false
9 | type-ratio-threshold: 0.5
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/liblinear.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.liblinear
2 |
3 | feat-index-file: src/main/resources/models/bioqa/answer_type/4b-dev-liblinear.findex
4 | label-index-file: src/main/resources/models/bioqa/answer_type/4b-dev-liblinear.lindex
5 | model-file: src/main/resources/models/bioqa/answer_type/4b-dev-liblinear.model
6 | solver-type: L1R_LR # L2R_L1LOSS_SVC_DUAL, L1R_LR, L2R_LR_DUAL, L2R_L2LOSS_SVC, L2R_L2LOSS_SVC_DUAL, L1R_L2LOSS_SVC, L2R_LR
7 | balance-weight: false
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/load-cv-liblinear-null.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.load-cv
2 |
3 | cv-predict-file: /models/bioqa/answer_type/4b-dev-liblinear-null.cv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/load-cv-liblinear.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.load-cv
2 |
3 | cv-predict-file: /models/bioqa/answer_type/4b-dev-liblinear.cv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/meka-pcc-null-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.predict
2 |
3 | classifier: 'inherit: bioqa.answer_type.meka-pcc-null'
4 | predict-file: result/answer-type-predict-meka-pcc-null.tsv
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/meka-pcc-null-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.train
2 |
3 | classifier: 'inherit: bioqa.answer_type.meka-pcc-null'
4 | cv-predict-file: src/main/resources/models/bioqa/answer_type/4b-dev-meka-pcc-null.cv
5 | at-gslabel-files:
6 | - /models/bioqa/answer_type/4b-dev-gslabel-uts.json
7 | - /models/bioqa/answer_type/4b-dev-gslabel-tmtool.json
8 | null-type: true
9 | type-ratio-threshold: 0.5
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/meka-pcc-null.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.meka-pcc
2 |
3 | model-file: src/main/resources/models/bioqa/answer_type/4b-dev-meka-pcc-null.model
4 | dataset-schema-file: src/main/resources/models/bioqa/answer_type/4b-dev-meka-pcc-null.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/meka-pcc-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.predict
2 |
3 | classifier: 'inherit: bioqa.answer_type.meka-pcc'
4 | predict-file: result/answer-type-predict-meka-pcc.tsv
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/meka-pcc-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.train
2 |
3 | classifier: 'inherit: bioqa.answer_type.meka-pcc'
4 | cv-predict-file: src/main/resources/models/bioqa/answer_type/4b-dev-meka-pcc.cv
5 | at-gslabel-files:
6 | - /models/bioqa/answer_type/4b-dev-gslabel-uts.json
7 | - /models/bioqa/answer_type/4b-dev-gslabel-tmtool.json
8 | null-type: false
9 | type-ratio-threshold: 0.5
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/meka-pcc.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.meka-pcc
2 |
3 | model-file: src/main/resources/models/bioqa/answer_type/4b-dev-meka-pcc.model
4 | dataset-schema-file: src/main/resources/models/bioqa/answer_type/4b-dev-meka-pcc.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/weka-knn-null-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.predict
2 |
3 | classifier: 'inherit: bioqa.answer_type.weka-knn-null'
4 | predict-file: result/answer-type-predict-weka-knn-null.tsv
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/weka-knn-null-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.train
2 |
3 | classifier: 'inherit: bioqa.answer_type.weka-knn-null'
4 | cv-predict-file: src/main/resources/models/bioqa/answer_type/4b-dev-weka-knn-null.cv
5 | at-gslabel-files:
6 | - /models/bioqa/answer_type/4b-dev-gslabel-uts.json
7 | - /models/bioqa/answer_type/4b-dev-gslabel-tmtool.json
8 | null-type: true
9 | type-ratio-threshold: 0.5
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/weka-knn-null.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka-knn
2 |
3 | model-file: src/main/resources/models/bioqa/answer_type/4b-dev-weka-knn-null.model
4 | dataset-schema-file: src/main/resources/models/bioqa/answer_type/4b-dev-weka-knn-null.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/weka-knn-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.predict
2 |
3 | classifier: 'inherit: bioqa.answer_type.weka-knn'
4 | predict-file: result/answer-type-predict-weka-knn.tsv
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/weka-knn-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.train
2 |
3 | classifier: 'inherit: bioqa.answer_type.weka-knn'
4 | cv-predict-file: src/main/resources/models/bioqa/answer_type/4b-dev-weka-knn.cv
5 | at-gslabel-files:
6 | - /models/bioqa/answer_type/4b-dev-gslabel-uts.json
7 | - /models/bioqa/answer_type/4b-dev-gslabel-tmtool.json
8 | null-type: false
9 | type-ratio-threshold: 0.5
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/weka-knn.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka-knn
2 |
3 | model-file: src/main/resources/models/bioqa/answer_type/4b-dev-weka-knn.model
4 | dataset-schema-file: src/main/resources/models/bioqa/answer_type/4b-dev-weka-knn.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/weka-reptree-null-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.predict
2 |
3 | classifier: 'inherit: bioqa.answer_type.weka-reptree-null'
4 | predict-file: result/answer-type-predict-weka-reptree-null.tsv
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/weka-reptree-null-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.train
2 |
3 | classifier: 'inherit: bioqa.answer_type.weka-reptree-null'
4 | cv-predict-file: src/main/resources/models/bioqa/answer_type/4b-dev-weka-reptree-null.cv
5 | at-gslabel-files:
6 | - /models/bioqa/answer_type/4b-dev-gslabel-uts.json
7 | - /models/bioqa/answer_type/4b-dev-gslabel-tmtool.json
8 | null-type: true
9 | type-ratio-threshold: 0.5
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/weka-reptree-null.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka-reptree
2 |
3 | model-file: src/main/resources/models/bioqa/answer_type/4b-dev-weka-reptree-null.model
4 | dataset-schema-file: src/main/resources/models/bioqa/answer_type/4b-dev-weka-reptree-null.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/weka-reptree-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.predict
2 |
3 | classifier: 'inherit: bioqa.answer_type.weka-reptree'
4 | predict-file: result/answer-type-predict-weka-reptree.tsv
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/weka-reptree-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.answer_type.train
2 |
3 | classifier: 'inherit: bioqa.answer_type.weka-reptree'
4 | cv-predict-file: src/main/resources/models/bioqa/answer_type/4b-dev-weka-reptree.cv
5 | at-gslabel-files:
6 | - /models/bioqa/answer_type/4b-dev-gslabel-uts.json
7 | - /models/bioqa/answer_type/4b-dev-gslabel-tmtool.json
8 | null-type: false
9 | type-ratio-threshold: 0.5
--------------------------------------------------------------------------------
/src/main/resources/bioqa/answer_type/weka-reptree.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka-reptree
2 |
3 | model-file: src/main/resources/models/bioqa/answer_type/4b-dev-weka-reptree.model
4 | dataset-schema-file: src/main/resources/models/bioqa/answer_type/4b-dev-weka-reptree.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/bioqa/concept/.gitignore:
--------------------------------------------------------------------------------
1 | rerank/scorers/lucene-bioconcept.yaml
2 | retrieval/lucene-bioconcept.yaml
--------------------------------------------------------------------------------
/src/main/resources/bioqa/concept/rerank/classifier-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.concept.rerank.classifier-predict
2 |
3 | scorers: |
4 | - inherit: baseqa.concept.rerank.scorers.original-score
5 | - inherit: baseqa.concept.rerank.scorers.name-match
6 | - inherit: bioqa.concept.rerank.scorers.lucene-bioconcept
--------------------------------------------------------------------------------
/src/main/resources/bioqa/concept/rerank/classifier-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.concept.rerank.classifier-train
2 |
3 | scorers: |
4 | - inherit: baseqa.concept.rerank.scorers.original-score
5 | - inherit: baseqa.concept.rerank.scorers.name-match
6 | - inherit: bioqa.concept.rerank.scorers.lucene-bioconcept
--------------------------------------------------------------------------------
/src/main/resources/bioqa/concept/rerank/liblinear-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.concept.rerank.classifier-predict
2 |
3 | classifier: 'inherit: bioqa.concept.rerank.liblinear'
4 | feature-file: result/concept-rerank-predict-liblinear.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/concept/rerank/liblinear-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.concept.rerank.classifier-train
2 |
3 | classifier: 'inherit: bioqa.concept.rerank.liblinear'
4 | cv-predict-file: src/main/resources/models/bioqa/concept/rerank/4b-dev-liblinear.cv
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/concept/rerank/liblinear.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.liblinear
2 |
3 | feat-index-file: src/main/resources/models/bioqa/concept/rerank/4b-dev-liblinear.findex
4 | label-index-file: src/main/resources/models/bioqa/concept/rerank/4b-dev-liblinear.lindex
5 | model-file: src/main/resources/models/bioqa/concept/rerank/4b-dev-liblinear.model
6 | solver-type: L1R_LR # L2R_L1LOSS_SVC_DUAL, L1R_LR, L2R_LR_DUAL, L2R_L2LOSS_SVC, L2R_L2LOSS_SVC_DUAL, L1R_L2LOSS_SVC, L2R_LR
7 | balance-weight: true
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/concept/rerank/scorers/lucene-bioconcept.yaml.template:
--------------------------------------------------------------------------------
1 | inherit: baseqa.concept.rerank.scorers.lucene
2 |
3 | fields: [name, synonym, definition]
4 | index: # ADD INDEX PATH HERE, E.G. ../index/bioconcept-lucene
5 | id-field: id
6 | source-field: source
7 | uri-prefix: /dictionaries/bioconcept-uri-prefix.tsv
8 | stoplist-path: /dictionaries/stoplist-gene-top5k.txt
--------------------------------------------------------------------------------
/src/main/resources/bioqa/concept/rerank/weka-logistic-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.concept.rerank.classifier-predict
2 |
3 | classifier: 'inherit: bioqa.concept.rerank.weka-logistic'
4 | feature-file: result/concept-rerank-predict-weka-logistic.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/concept/rerank/weka-logistic-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.concept.rerank.classifier-train
2 |
3 | classifier: 'inherit: bioqa.concept.rerank.weka-logistic'
4 | cv-predict-file: src/main/resources/models/bioqa/concept/rerank/4b-dev-weka-logistic.cv
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/concept/rerank/weka-logistic.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka-logistic
2 |
3 | model-file: src/main/resources/models/bioqa/concept/rerank/4b-dev-weka-logistic.model
4 | dataset-schema-file: src/main/resources/models/bioqa/concept/rerank/4b-dev-weka-logistic.dataset-schema
5 | dataset-export: src/main/resources/models/bioqa/concept/rerank/4b-dev-weka.arff
6 | balance-weight: false
7 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/concept/retrieval/lucene-bioconcept.yaml.template:
--------------------------------------------------------------------------------
1 | inherit: baseqa.concept.retrieval.lucene
2 |
3 | fields: [name, synonym, definition]
4 | index: # ADD INDEX PATH HERE, E.G. ../index/bioconcept-lucene
5 | id-field: id
6 | name-field: name
7 | source-field: source
8 | uri-prefix: /dictionaries/bioconcept-uri-prefix.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/.gitignore:
--------------------------------------------------------------------------------
1 | rerank/scorers/lucene-medline.yaml
2 | retrieval/lucene-medline.yaml
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/rerank/classifier-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.document.rerank.classifier-predict
2 |
3 | scorers: |
4 | - inherit: baseqa.document.rerank.scorers.original-score
5 | - inherit: bioqa.document.rerank.scorers.lucene-medline
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/rerank/classifier-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.document.rerank.classifier-train
2 |
3 | scorers: |
4 | - inherit: baseqa.document.rerank.scorers.original-score
5 | - inherit: bioqa.document.rerank.scorers.lucene-medline
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/rerank/cv-load-liblinear.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.document.rerank.cv-load
2 |
3 | cv-predict-file: /models/bioqa/document/rerank/4b-dev-liblinear.cv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/rerank/liblinear-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.document.rerank.classifier-predict
2 |
3 | classifier: 'inherit: bioqa.document.rerank.liblinear'
4 | feature-file: result/document-rerank-predict-liblinear.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/rerank/liblinear-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.document.rerank.classifier-train
2 |
3 | classifier: 'inherit: bioqa.document.rerank.liblinear'
4 | cv-predict-file: src/main/resources/models/bioqa/document/rerank/4b-dev-liblinear.cv
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/rerank/liblinear.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.liblinear
2 |
3 | feat-index-file: src/main/resources/models/bioqa/document/rerank/4b-dev-liblinear.findex
4 | label-index-file: src/main/resources/models/bioqa/document/rerank/4b-dev-liblinear.lindex
5 | model-file: src/main/resources/models/bioqa/document/rerank/4b-dev-liblinear.model
6 | solver-type: L1R_LR # L2R_L1LOSS_SVC_DUAL, L1R_LR, L2R_LR_DUAL, L2R_L2LOSS_SVC, L2R_L2LOSS_SVC_DUAL, L1R_L2LOSS_SVC, L2R_LR
7 | balance-weight: true
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/rerank/logreg-params.txt:
--------------------------------------------------------------------------------
1 | -3.5976 -0.0115 0.2500 0.2350
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/rerank/logreg-pubmed.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.document.rerank.logreg
2 |
3 | query-string-contructor: edu.cmu.lti.oaqa.bioqa.providers.query.PubMedQueryStringConstructor
4 | doc-logreg-params: /bioqa/document/rerank/logreg-params.txt
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/rerank/scorers/lucene-medline.yaml.template:
--------------------------------------------------------------------------------
1 | inherit: baseqa.document.rerank.scorers.lucene
2 |
3 | fields: [articleTitle, abstractText]
4 | index: # ADD IDNEX PATH HERE, E.G. ../index/medline16n-lucene
5 | id-field: pmid
6 | uri-prefix: http://www.ncbi.nlm.nih.gov/pubmed/
7 | stoplist-path: /dictionaries/stoplist-gene-top5k.txt
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/rerank/weka-logistic-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.document.rerank.classifier-predict
2 |
3 | classifier: 'inherit: bioqa.document.rerank.weka-logistic'
4 | feature-file: result/document-rerank-predict-weka-logistic.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/rerank/weka-logistic-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.document.rerank.classifier-train
2 |
3 | classifier: 'inherit: bioqa.document.rerank.weka-logistic'
4 | cv-predict-file: src/main/resources/models/bioqa/document/rerank/4b-dev-weka-logistic.cv
5 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/rerank/weka-logistic.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.weka-logistic
2 |
3 | model-file: src/main/resources/models/bioqa/document/rerank/4b-dev-weka-logistic.model
4 | dataset-schema-file: src/main/resources/models/bioqa/document/rerank/4b-dev-weka-logistic.dataset-schema
5 | dataset-export: src/main/resources/models/bioqa/document/rerank/4b-dev-weka.arff
6 | balance-weight: false
7 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/document/retrieval/lucene-medline.yaml.template:
--------------------------------------------------------------------------------
1 | inherit: baseqa.document.retrieval.lucene
2 |
3 | fields: [articleTitle, abstractText]
4 | index: # ADD IDNEX PATH HERE, E.G. ../index/medline16n-lucene
5 | id-field: pmid
6 | title-field: articleTitle
7 | text-field: abstractText
8 | uri-prefix: http://www.ncbi.nlm.nih.gov/pubmed/
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/concept/lingpipe-genetag.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.concept.PassageConceptRecognizer
2 |
3 | concept-provider: 'inherit: bioqa.providers.kb.lingpipe-genetag'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/concept/lingpipe-genia.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.concept.PassageConceptRecognizer
2 |
3 | concept-provider: 'inherit: bioqa.providers.kb.lingpipe-genia'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/concept/metamap-cached.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.concept.PassageConceptRecognizer
2 |
3 | allowed-concept-types: /dictionaries/allowed-umls-types.txt
4 | concept-provider: 'inherit: bioqa.providers.kb.metamap-cached'
5 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/concept/search-uts-cached.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.evidence.concept.search
2 |
3 | concept-search-provider: 'inherit: bioqa.providers.kb.concept-search-uts-cached'
4 | synonym-expansion-provider: 'inherit: bioqa.providers.kb.synonym-uts-cached'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/concept/search-uts.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.evidence.concept.search
2 |
3 | concept-search-provider: 'inherit: bioqa.providers.kb.concept-search-uts'
4 | synonym-expansion-provider: 'inherit: bioqa.providers.kb.synonym-uts'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/concept/tmtool-cached.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.concept.PassageConceptRecognizer
2 |
3 | concept-provider: 'inherit: bioqa.providers.kb.tmtool-cached'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/concept/tmtool.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.concept.PassageConceptRecognizer
2 |
3 | concept-provider: 'inherit: bioqa.providers.kb.tmtool'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/parse/clearnlp-bioinformatics.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.parse.PassageParser
2 |
3 | parser-provider: 'inherit: bioqa.providers.parser.clearnlp-bioinformatics'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/parse/clearnlp-medical.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.parse.PassageParser
2 |
3 | parser-provider: 'inherit: bioqa.providers.parser.clearnlp-medical'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/parse/lingpipe-indoeuro-genia.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.parse.PassageParser
2 |
3 | parser-provider: 'inherit: bioqa.providers.parser.lingpipe-indoeuro-genia'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/parse/lingpipe-indoeuro-medpost.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.parse.PassageParser
2 |
3 | parser-provider: 'inherit: bioqa.providers.parser.lingpipe-indoeuro-medpost'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/parse/lingpipe-regex-genia.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.parse.PassageParser
2 |
3 | parser-provider: 'inherit: bioqa.providers.parser.lingpipe-regex-genia'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/evidence/parse/lingpipe-regex-medpost.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.evidence.parse.PassageParser
2 |
3 | parser-provider: 'inherit: bioqa.providers.parser.lingpipe-regex-medpost'
4 | view-name-prefix: ptv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/passage/rerank/classifier-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.passage.rerank.classifier-predict
2 |
3 | scorers: |
4 | - inherit: baseqa.passage.rerank.scorers.original-score
5 | - inherit: baseqa.passage.rerank.scorers.meta-info
6 | - inherit: bioqa.passage.rerank.scorers.lucene
7 | # add "classifier", "feature-file"
--------------------------------------------------------------------------------
/src/main/resources/bioqa/passage/rerank/classifier-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.passage.rerank.classifier-train
2 |
3 | scorers: |
4 | - inherit: baseqa.passage.rerank.scorers.original-score
5 | - inherit: baseqa.passage.rerank.scorers.meta-info
6 | - inherit: bioqa.passage.rerank.scorers.lucene
7 | resample-type: DOWN
8 | # add "classifier", "cv-predict-file"
--------------------------------------------------------------------------------
/src/main/resources/bioqa/passage/rerank/liblinear-predict.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.passage.rerank.classifier-predict
2 |
3 | classifier: 'inherit: bioqa.passage.rerank.liblinear'
4 | feature-file: result/passage-rerank-predict-liblinear.tsv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/passage/rerank/liblinear-train.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.passage.rerank.classifier-train
2 |
3 | classifier: 'inherit: bioqa.passage.rerank.liblinear'
4 | cv-predict-file: src/main/resources/models/bioqa/passage/rerank/4b-dev-liblinear.cv
--------------------------------------------------------------------------------
/src/main/resources/bioqa/passage/rerank/liblinear.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.providers.ml.classifiers.liblinear
2 |
3 | feat-index-file: src/main/resources/models/bioqa/passage/rerank/4b-dev-liblinear.findex
4 | label-index-file: src/main/resources/models/bioqa/passage/rerank/4b-dev-liblinear.lindex
5 | model-file: src/main/resources/models/bioqa/passage/rerank/4b-dev-liblinear.model
6 | solver-type: L1R_LR # L2R_L1LOSS_SVC_DUAL, L1R_LR, L2R_LR_DUAL, L2R_L2LOSS_SVC, L2R_L2LOSS_SVC_DUAL, L1R_L2LOSS_SVC, L2R_LR
7 | balance-weight: true
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/passage/rerank/scorers/lucene.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.passage.rerank.scorers.lucene
2 |
3 | stoplist-path: /dictionaries/stoplist-gene-top5k.txt
--------------------------------------------------------------------------------
/src/main/resources/bioqa/passage/retrieval/lucene-sentence-medline-improve.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.passage.retrieval.lucene-sentence-improve
2 |
3 | sentence-model: com.aliasi.sentences.MedlineSentenceModel
4 | parser-provider: 'inherit: bioqa.providers.parser.clearnlp-bioinformatics'
5 | hits: 100
6 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/passage/retrieval/lucene-sentence-medline.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.passage.retrieval.lucene-sentence
2 |
3 | sentence-model: com.aliasi.sentences.MedlineSentenceModel
4 | parser-provider: 'inherit: bioqa.providers.parser.clearnlp-bioinformatics'
5 | hits: 100
6 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/preprocess/passage-kb-cache.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.preprocess.passage-concept-cache
2 |
3 | batch-size: 100
4 | concept-providers: |
5 | - inherit: bioqa.providers.kb.tmtool-cached
6 | - inherit: bioqa.providers.kb.metamap-cached
7 | synonym-expansion-providers: |
8 | - inherit: bioqa.providers.kb.synonym-uts-cached
9 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/preprocess/question-kb-cache.yaml:
--------------------------------------------------------------------------------
1 | inherit: baseqa.preprocess.question-concept-cache
2 |
3 | concept-providers: |
4 | - inherit: bioqa.providers.kb.tmtool-cached
5 | - inherit: bioqa.providers.kb.metamap-cached
6 | synonym-expansion-providers: |
7 | - inherit: bioqa.providers.kb.synonym-uts-cached
8 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/kb/.gitignore:
--------------------------------------------------------------------------------
1 | concept-search-uts.yaml
2 | concept-search-uts-cached.yaml
3 | metamap.yaml
4 | metamap-cached.yaml
5 | synonym-uts.yaml
6 | synonym-uts-cached.yaml
7 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/kb/concept-search-uts-cached.yaml.template:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioqa.providers.kb.CachedUtsConceptSearchProvider
2 |
3 | service: http://umlsks.nlm.nih.gov
4 | version: # ADD VERSION HERE, E.G. 2015AB
5 | username: # ADD USERNAME HERE
6 | password: # ADD PASSWORD HERE
7 |
8 | db-file: src/main/resources/concept-search-cache/uts-cache.mapdb
9 | map-name: uts-cache
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/kb/concept-search-uts.yaml.template:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioqa.providers.kb.UtsConceptSearchProvider
2 |
3 | service: http://umlsks.nlm.nih.gov
4 | version: # ADD VERSION HERE, E.G. 2015AB
5 | username: # ADD USERNAME HERE
6 | password: # ADD PASSWORD HERE
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/kb/lingpipe-genetag.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.kb.LingPipeNerConceptProvider
2 |
3 | chunker-model: /ne-en-bio-genetag.HmmChunker
4 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/kb/lingpipe-genia.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.kb.LingPipeNerConceptProvider
2 |
3 | chunker-model: /ne-en-bio-genia.TokenShapeChunker
4 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/kb/metamap-cached.yaml.template:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioqa.providers.kb.CachedMetaMapConceptProvider
2 |
3 | version: # ADD VERSION HERE, E.G. '1516'
4 | username: # ADD USERNAME HERE
5 | password: # ADD PASSWORD HERE
6 | email: # ADD EMAIL ADD
7 |
8 | db-file: src/main/resources/metamap-cache/metamap-cache.mapdb
9 | map-name: metamap-cache
10 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/kb/metamap.yaml.template:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioqa.providers.kb.MetaMapConceptProvider
2 |
3 | version: # ADD VERSION HERE, E.G. '1516'
4 | username: # ADD USERNAME HERE
5 | password: # ADD PASSWORD HERE
6 | email: # ADD EMAIL ADD
7 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/kb/synonym-uts-cached.yaml.template:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioqa.providers.kb.CachedUtsSynonymExpansionProvider
2 |
3 | service: http://umlsks.nlm.nih.gov
4 | version: # ADD VERSION HERE, E.G. 2015AB
5 | username: # ADD USERNAME HERE
6 | password: # ADD PASSWORD HERE
7 | nthreads: 100
8 | timeout: 5
9 |
10 | db-file: src/main/resources/synonym-cache/uts-cache.mapdb
11 | map-name: uts-cache
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/kb/synonym-uts.yaml.template:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioqa.providers.kb.UtsSynonymExpansionProvider
2 |
3 | service: http://umlsks.nlm.nih.gov
4 | version: # ADD VERSION HERE, E.G. 2015AB
5 | username: # ADD USERNAME HERE
6 | password: # ADD PASSWORD HERE
7 | nthreads: 100
8 | timeout: 5
9 |
10 | db-file: src/main/resources/synonym-cache/uts-cache.mapdb
11 | map-name: uts-cache
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/kb/tmtool-cached.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioqa.providers.kb.CachedTmToolConceptProvider
2 |
3 | triggers: [tmChem, DNorm, tmVar, GNormPlus]
4 |
5 | db-file: src/main/resources/tmtool-cache/tmtool-cache.mapdb
6 | map-name: tmtool-cache
7 |
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/kb/tmtool.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.bioqa.providers.kb.TmToolConceptProvider
2 |
3 | triggers: [tmChem, DNorm, tmVar, GNormPlus]
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/parser/clearnlp-bioinformatics.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.parser.ClearParserProvider
2 |
3 | pos-model: /bioinformatics-en-pos.xz
4 | dep-model: /bioinformatics-en-dep.xz
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/parser/clearnlp-medical.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.parser.ClearParserProvider
2 |
3 | pos-model: /medical-en-pos.xz
4 | dep-model: /medical-en-dep.xz
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/parser/lingpipe-indoeuro-genia.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.parser.LingPipeParserProvider
2 |
3 | token-factory: com.aliasi.tokenizer.IndoEuropeanTokenizerFactory
4 | token-factory-params: []
5 | pos-model: /pos-en-bio-genia.HiddenMarkovModel
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/parser/lingpipe-indoeuro-medpost.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.parser.LingPipeParserProvider
2 |
3 | token-factory: com.aliasi.tokenizer.IndoEuropeanTokenizerFactory
4 | token-factory-params: []
5 | pos-model: /pos-en-bio-medpost.HiddenMarkovModel
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/parser/lingpipe-regex-genia.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.parser.LingPipeParserProvider
2 |
3 | token-factory: com.aliasi.tokenizer.RegExTokenizerFactory
4 | token-factory-params: ['([-''\d\p{L}]|\([-''\d\p{L}]+\))*[-''\d\p{L}]+|\S']
5 | pos-model: /pos-en-bio-genia.HiddenMarkovModel
--------------------------------------------------------------------------------
/src/main/resources/bioqa/providers/parser/lingpipe-regex-medpost.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.providers.parser.LingPipeParserProvider
2 |
3 | token-factory: com.aliasi.tokenizer.RegExTokenizerFactory
4 | token-factory-params: ['([-''\d\p{L}]|\([-''\d\p{L}]+\))*[-''\d\p{L}]+|\S']
5 | pos-model: /pos-en-bio-medpost.HiddenMarkovModel
--------------------------------------------------------------------------------
/src/main/resources/bioqa/question/concept/lingpipe-genetag.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.concept.QuestionConceptRecognizer
2 |
3 | concept-provider: 'inherit: bioqa.providers.kb.lingpipe-genetag'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/question/concept/lingpipe-genia.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.concept.QuestionConceptRecognizer
2 |
3 | concept-provider: 'inherit: bioqa.providers.kb.lingpipe-genia'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/question/concept/metamap-cached.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.question.concept.metamap
2 |
3 | concept-provider: 'inherit: bioqa.providers.kb.metamap-cached'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/question/concept/metamap.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.concept.QuestionConceptRecognizer
2 |
3 | concept-provider: 'inherit: bioqa.providers.kb.metamap'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/question/concept/tmtool-cached.yaml:
--------------------------------------------------------------------------------
1 | inherit: bioqa.question.concept.tmtool
2 |
3 | concept-provider: 'inherit: bioqa.providers.kb.tmtool-cached'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/question/concept/tmtool.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.concept.QuestionConceptRecognizer
2 |
3 | concept-provider: 'inherit: bioqa.providers.kb.tmtool'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/question/parse/clearnlp-bioinformatics.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.parse.QuestionParser
2 |
3 | parser-provider: 'inherit: bioqa.providers.parser.clearnlp-bioinformatics'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/question/parse/clearnlp-medical.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.parse.QuestionParser
2 |
3 | parser-provider: 'inherit: bioqa.providers.parser.clearnlp-medical'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/question/parse/lingpipe-indoeuro-genia.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.parse.QuestionParser
2 |
3 | parser-provider: 'inherit: bioqa.providers.parser.lingpipe-indoeuro-genia'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/question/parse/lingpipe-indoeuro-medpost.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.parse.QuestionParser
2 |
3 | parser-provider: 'inherit: bioqa.providers.parser.lingpipe-indoeuro-medpost'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/question/parse/lingpipe-regex-genia.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.parse.QuestionParser
2 |
3 | parser-provider: 'inherit: bioqa.providers.parser.lingpipe-regex-genia'
--------------------------------------------------------------------------------
/src/main/resources/bioqa/question/parse/lingpipe-regex-medpost.yaml:
--------------------------------------------------------------------------------
1 | class: edu.cmu.lti.oaqa.baseqa.question.parse.QuestionParser
2 |
3 | parser-provider: 'inherit: bioqa.providers.parser.lingpipe-regex-medpost'
--------------------------------------------------------------------------------
/src/main/resources/dictionaries/allowed-umls-types.txt:
--------------------------------------------------------------------------------
1 | umls:aapp
2 | umls:amas
3 | umls:anab
4 | umls:antb
5 | umls:bacs
6 | umls:bact
7 | umls:bdsu
8 | umls:bird
9 | umls:blor
10 | umls:bmod
11 | umls:bpoc
12 | umls:bsoj
13 | umls:carb
14 | umls:celc
15 | umls:celf
16 | umls:cell
17 | umls:cgab
18 | umls:clas
19 | umls:clna
20 | umls:clnd
21 | umls:comd
22 | umls:diap
23 | umls:dora
24 | umls:dsyn
25 | umls:elii
26 | umls:emod
27 | umls:enzy
28 | umls:euka
29 | umls:fish
30 | umls:fndg
31 | umls:fngs
32 | umls:food
33 | umls:ftcn
34 | umls:genf
35 | umls:geoa
36 | umls:gngm
37 | umls:hlca
38 | umls:hops
39 | umls:horm
40 | umls:idcn
41 | umls:imft
42 | umls:inbe
43 | umls:inch
44 | umls:inpr
45 | umls:irda
46 | umls:lbpr
47 | umls:lipd
48 | umls:mamm
49 | umls:mbrt
50 | umls:medd
51 | umls:menp
52 | umls:mobd
53 | umls:moft
54 | umls:neop
55 | umls:nnon
56 | umls:nsba
57 | umls:opco
58 | umls:orch
59 | umls:orga
60 | umls:orgf
61 | umls:ortf
62 | umls:patf
63 | umls:phpr
64 | umls:phsf
65 | umls:phsu
66 | umls:plnt
67 | umls:popg
68 | umls:qlco
69 | umls:qnco
70 | umls:rcpt
71 | umls:resa
72 | umls:resd
73 | umls:sbst
74 | umls:sosy
75 | umls:spco
76 | umls:strd'
77 | umls:tisu
78 | umls:tmco
79 | umls:topp
80 | umls:virs
81 | umls:vita
--------------------------------------------------------------------------------
/src/main/resources/dictionaries/bioconcept-uri-prefix.tsv:
--------------------------------------------------------------------------------
1 | DISEASE http://www.disease-ontology.org/api/metadata/
2 | GENE http://amigo.geneontology.org/cgi-bin/amigo/term_details?term=
3 | JOCHEM http://www.biosemantics.org/jochem#
4 | MESH http://www.nlm.nih.gov/cgi/mesh/2016/MB_cgi?field=uid&exact=Find+Exact+Term&term=
5 | UNIPROT http://www.uniprot.org/uniprot/
--------------------------------------------------------------------------------
/src/main/resources/dictionaries/negation-cues.txt:
--------------------------------------------------------------------------------
1 | aint
2 | cannot
3 | cant
4 | darent
5 | didnt
6 | doesnt
7 | dont
8 | hadnt
9 | hardly
10 | hasnt
11 | havent
12 | havnt
13 | isnt
14 | lack
15 | lacking
16 | lacks
17 | mightnt
18 | mustnt
19 | neednt
20 | neither
21 | never
22 | no
23 | nobody
24 | none
25 | nor
26 | not
27 | nothing
28 | nowhere
29 | n’t
30 | oughtnt
31 | shant
32 | shouldnt
33 | wasnt
34 | without
35 | wouldnt
--------------------------------------------------------------------------------
/src/main/resources/dictionaries/noun-tags.txt:
--------------------------------------------------------------------------------
1 | NN
2 | NNS
3 | NNP
4 | NNPS
--------------------------------------------------------------------------------
/src/main/resources/dictionaries/pos-tags.txt:
--------------------------------------------------------------------------------
1 | FW
2 | JJ
3 | JJR
4 | JJS
5 | NN
6 | NNS
7 | NNP
8 | NNPS
9 | VB
10 | VBD
11 | VBG
12 | VBN
13 | VBP
14 | VBZ
--------------------------------------------------------------------------------
/src/main/resources/dictionaries/quantity-question-words.txt:
--------------------------------------------------------------------------------
1 | how many
2 | how much
3 | how large
4 | how long
5 | diameter
6 | value
7 | rate
8 | percentage
9 | incidence
10 | prevalence
11 | proportion
12 | number
--------------------------------------------------------------------------------
/src/main/resources/dictionaries/stoplist-gene-modified.txt:
--------------------------------------------------------------------------------
1 | a
2 | about
3 | again
4 | all
5 | almost
6 | also
7 | although
8 | always
9 | among
10 | an
11 | and
12 | another
13 | any
14 | are
15 | as
16 | at
17 | be
18 | because
19 | been
20 | before
21 | being
22 | between
23 | both
24 | but
25 | by
26 | can
27 | could
28 | did
29 | do
30 | does
31 | done
32 | due
33 | during
34 | each
35 | either
36 | enough
37 | especially
38 | etc
39 | for
40 | found
41 | from
42 | further
43 | gene
44 | genes
45 | genome
46 | genomes
47 | had
48 | has
49 | have
50 | having
51 | here
52 | how
53 | however
54 | i
55 | if
56 | in
57 | into
58 | is
59 | it
60 | its
61 | itself
62 | just
63 | kg
64 | km
65 | made
66 | mainly
67 | make
68 | may
69 | mg
70 | might
71 | ml
72 | mm
73 | most
74 | mostly
75 | must
76 | nearly
77 | neither
78 | no
79 | nor
80 | obtained
81 | of
82 | often
83 | on
84 | our
85 | overall
86 | perhaps
87 | protein
88 | quite
89 | rather
90 | really
91 | regarding
92 | seem
93 | seen
94 | sequence
95 | several
96 | should
97 | show
98 | showed
99 | shown
100 | shows
101 | significantly
102 | since
103 | so
104 | some
105 | such
106 | than
107 | that
108 | the
109 | their
110 | theirs
111 | them
112 | then
113 | there
114 | therefore
115 | these
116 | they
117 | this
118 | those
119 | through
120 | thus
121 | to
122 | upon
123 | use
124 | used
125 | using
126 | various
127 | very
128 | was
129 | we
130 | were
131 | what
132 | when
133 | which
134 | while
135 | with
136 | within
137 | without
138 | would
--------------------------------------------------------------------------------
/src/main/resources/dictionaries/stoplist-gene.txt:
--------------------------------------------------------------------------------
1 | a
2 | about
3 | again
4 | all
5 | almost
6 | also
7 | although
8 | always
9 | among
10 | an
11 | and
12 | another
13 | any
14 | are
15 | as
16 | at
17 | be
18 | because
19 | been
20 | before
21 | being
22 | between
23 | both
24 | but
25 | by
26 | can
27 | could
28 | did
29 | do
30 | does
31 | done
32 | due
33 | during
34 | each
35 | either
36 | enough
37 | especially
38 | etc
39 | for
40 | found
41 | from
42 | further
43 | gene
44 | had
45 | has
46 | have
47 | having
48 | here
49 | how
50 | however
51 | i
52 | if
53 | in
54 | into
55 | is
56 | it
57 | its
58 | itself
59 | just
60 | kg
61 | km
62 | made
63 | mainly
64 | make
65 | may
66 | mg
67 | might
68 | ml
69 | mm
70 | most
71 | mostly
72 | must
73 | nearly
74 | neither
75 | no
76 | nor
77 | obtained
78 | of
79 | often
80 | on
81 | our
82 | overall
83 | perhaps
84 | protein
85 | quite
86 | rather
87 | really
88 | regarding
89 | seem
90 | seen
91 | sequence
92 | several
93 | should
94 | show
95 | showed
96 | shown
97 | shows
98 | significantly
99 | since
100 | so
101 | some
102 | such
103 | than
104 | that
105 | the
106 | their
107 | theirs
108 | them
109 | then
110 | there
111 | therefore
112 | these
113 | they
114 | this
115 | those
116 | through
117 | thus
118 | to
119 | upon
120 | use
121 | used
122 | using
123 | various
124 | very
125 | was
126 | we
127 | were
128 | what
129 | when
130 | which
131 | while
132 | with
133 | within
134 | without
135 | would
--------------------------------------------------------------------------------
/src/main/resources/dictionaries/stoplist.txt:
--------------------------------------------------------------------------------
1 | a
2 | about
3 | again
4 | all
5 | almost
6 | also
7 | although
8 | always
9 | among
10 | an
11 | and
12 | another
13 | any
14 | are
15 | as
16 | at
17 | be
18 | because
19 | been
20 | before
21 | being
22 | between
23 | both
24 | but
25 | by
26 | can
27 | could
28 | did
29 | do
30 | does
31 | done
32 | due
33 | during
34 | each
35 | either
36 | enough
37 | especially
38 | etc
39 | for
40 | found
41 | from
42 | further
43 | gene
44 | had
45 | has
46 | have
47 | having
48 | here
49 | how
50 | however
51 | i
52 | if
53 | in
54 | into
55 | is
56 | it
57 | its
58 | itself
59 | just
60 | kg
61 | km
62 | made
63 | mainly
64 | make
65 | may
66 | mg
67 | might
68 | ml
69 | mm
70 | most
71 | mostly
72 | must
73 | nearly
74 | neither
75 | no
76 | nor
77 | obtained
78 | of
79 | often
80 | on
81 | our
82 | overall
83 | perhaps
84 | protein
85 | quite
86 | rather
87 | really
88 | regarding
89 | seem
90 | seen
91 | sequence
92 | several
93 | should
94 | show
95 | showed
96 | shown
97 | shows
98 | significantly
99 | since
100 | so
101 | some
102 | such
103 | than
104 | that
105 | the
106 | their
107 | theirs
108 | them
109 | then
110 | there
111 | therefore
112 | these
113 | they
114 | this
115 | those
116 | through
117 | thus
118 | to
119 | upon
120 | use
121 | used
122 | using
123 | various
124 | very
125 | was
126 | we
127 | were
128 | what
129 | when
130 | which
131 | while
132 | with
133 | within
134 | without
135 | would
--------------------------------------------------------------------------------
/src/main/resources/log4j.properties:
--------------------------------------------------------------------------------
1 | #
2 | # Open Advancement Question Answering (OAQA) Project Copyright 2017 Carnegie Mellon University
3 | #
4 | # Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
5 | # in compliance with the License. You may obtain a copy of the License at
6 | #
7 | # http://www.apache.org/licenses/LICENSE-2.0
8 | #
9 | # Unless required by applicable law or agreed to in writing, software distributed under the License
10 | # is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
11 | # or implied. See the License for the specific language governing permissions and limitations
12 | # under the License.
13 | #
14 |
15 | log4j.appender.stdout=org.apache.log4j.ConsoleAppender
16 | log4j.appender.stdout.Target=System.out
17 | log4j.appender.stdout.layout=org.apache.log4j.PatternLayout
18 | log4j.appender.stdout.layout.ConversionPattern=[%p] [%c{1}] %m%n
19 | log4j.logger.edu.cmu.lti.oaqa=DEBUG, stdout
--------------------------------------------------------------------------------
/src/main/resources/models/bioasq/concept/rerank/4b-dev-liblinear.findex:
--------------------------------------------------------------------------------
1 | 1 tokens_concatenated@MESH/rank
2 | 2 cmention_names_concatenated@UNIPROT/score
3 | 3 w_id_concept_names_individual@JOCHEM/score
4 | 4 cmention_names_individual@DISEASE/rank
5 | 5 concept_names_concatenated@UNIPROT/score
6 | 6 tokens_concatenated@DISEASE/rank
7 | 7 cmention_names_individual@JOCHEM/rank
8 | 8 w_id_concept_names_individual@DISEASE/score
9 | 9 tokens_concatenated@GENE/rank
10 | 10 concept_names_concatenated@DISEASE/rank
11 | 11 w_id_concept_names_individual@UNIPROT/rank
12 | 12 cmention_names_concatenated@MESH/rank
13 | 13 concept_names_concatenated@GENE/rank
14 | 14 cmention_names_individual@UNIPROT/score
15 | 15 cmention_names_concatenated@DISEASE/rank
16 | 16 concept_names_concatenated@MESH/score
17 | 17 w_id_concept_names_individual@UNIPROT/score
18 | 18 tokens_concatenated@JOCHEM/score
19 | 19 cmention_names_individual@UNIPROT/rank
20 | 20 cmention_names_concatenated@GENE/rank
21 | 21 w_id_concept_names_individual@MESH/score
22 | 22 concept-name-match
23 | 23 original/score
24 | 24 original/rank
25 | 25 concept_names_concatenated@JOCHEM/score
26 | 26 cmention_names_concatenated@GENE/score
27 | 27 concept_names_concatenated@DISEASE/score
28 | 28 cmention_names_individual@JOCHEM/score
29 | 29 tokens_concatenated@GENE/score
30 | 30 cmention_names_concatenated@DISEASE/score
31 | 31 w_id_concept_names_individual@MESH/rank
32 | 32 cmention_names_individual@MESH/score
33 | 33 concept_names_concatenated@JOCHEM/rank
34 | 34 cmention_names_individual@MESH/rank
35 | 35 cmention_names_individual@DISEASE/score
36 | 36 concept_names_concatenated@UNIPROT/rank
37 | 37 tokens_concatenated@UNIPROT/rank
38 | 38 w_id_concept_names_individual@JOCHEM/rank
39 | 39 cmention_names_concatenated@MESH/score
40 | 40 w_id_concept_names_individual@GENE/rank
41 | 41 cmention_names_individual@GENE/rank
42 | 42 concept_names_concatenated@GENE/score
43 | 43 w_id_concept_names_individual@GENE/score
44 | 44 cmention_names_concatenated@UNIPROT/rank
45 | 45 cmention_names_concatenated@JOCHEM/rank
46 | 46 w_id_concept_names_individual@DISEASE/rank
47 | 47 cmention-name-match
48 | 48 tokens_concatenated@JOCHEM/rank
49 | 49 tokens_concatenated@DISEASE/score
50 | 50 cmention_names_individual@GENE/score
51 | 51 concept_names_concatenated@MESH/rank
52 | 52 tokens_concatenated@MESH/score
53 | 53 tokens_concatenated@UNIPROT/score
54 | 54 cmention_names_concatenated@JOCHEM/score
--------------------------------------------------------------------------------
/src/main/resources/models/bioasq/concept/rerank/4b-dev-liblinear.lindex:
--------------------------------------------------------------------------------
1 | 2 true
2 | 1 false
--------------------------------------------------------------------------------
/src/main/resources/models/bioasq/concept/rerank/4b-dev-liblinear.model:
--------------------------------------------------------------------------------
1 | solver_type L1R_LR
2 | nr_class 2
3 | label 1 2
4 | nr_feature 53
5 | bias 0.000000000000000
6 | w
7 | -0.4623546270993043
8 | 0
9 | 0.3994119587204691
10 | -1.569117167451501
11 | 0.3114856122461142
12 | 0
13 | 0
14 | -0.02279194141608566
15 | 0
16 | 0
17 | 0
18 | 0
19 | 0
20 | -0.1760443750805484
21 | 0
22 | 0.1732120675784405
23 | 0
24 | -1.113047689458425
25 | 0
26 | 0
27 | -0.1233853912498214
28 | -1.891144054938629
29 | 0.08844002407650575
30 | -5.226937073850687
31 | 0.2617539510046160
32 | 0
33 | 0
34 | 0.006884867429180922
35 | 0
36 | -0.7042847154175214
37 | 0
38 | 0.4368007810015496
39 | 0
40 | -3.075296977163223
41 | 0.3056294515482870
42 | 0
43 | 0
44 | 0
45 | -0.2564862488899015
46 | 0
47 | 0
48 | 0
49 | 4.056526829336180
50 | 0
51 | 0
52 | 0.5563161839949544
53 | -0.8041624243772714
54 | 0
55 | -0.4678671582635070
56 | 0
57 | 0
58 | -0.7151584713292598
59 | -1.376197824893477
60 | 0
61 |
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/collective_score/4b-dev-liblinear.lindex:
--------------------------------------------------------------------------------
1 | 2 true
2 | 1 false
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/collective_score/4b-dev-weka-cvr.dataset-schema:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/answer/collective_score/4b-dev-weka-cvr.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/collective_score/4b-dev-weka-cvr.model:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/answer/collective_score/4b-dev-weka-cvr.model
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/score/4b-dev-liblinear.lindex:
--------------------------------------------------------------------------------
1 | 2 false
2 | 1 true
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/score/4b-dev-weka-cvr.dataset-schema:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/answer/score/4b-dev-weka-cvr.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/score/4b-dev-weka-cvr.model:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/answer/score/4b-dev-weka-cvr.model
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/yesno/4b-dev-liblinear.lindex:
--------------------------------------------------------------------------------
1 | 2 no
2 | 1 yes
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-cvr.dataset-schema:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-cvr.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-cvr.model:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-cvr.model
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-logistic.dataset-schema:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-logistic.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-logistic.model:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-logistic.model
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-other.dataset-schema:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-other.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-other.model:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/answer/yesno/4b-dev-weka-other.model
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer_type/4b-dev-liblinear-null.lindex:
--------------------------------------------------------------------------------
1 | 17 umls:gngm
2 | 40 umls:lbpr
3 | 23 umls:virs
4 | 46 tmtool:SNP
5 | 9 umls:bpoc
6 | 15 umls:dsyn
7 | 32 tmtool:Chemical
8 | 55 umls:inpo
9 | 38 umls:clas
10 | 7 umls:aapp
11 | 13 umls:fngs
12 | 30 umls:genf
13 | 53 umls:ftcn
14 | 36 umls:sosy
15 | 59 umls:orga
16 | 5 umls:fndg
17 | 22 umls:comd
18 | 45 umls:spco
19 | 28 umls:celf
20 | 51 umls:cgab
21 | 34 tmtool:Disease
22 | 57 tmtool:Species
23 | 3 null
24 | 20 umls:moft
25 | 43 umls:antb
26 | 26 umls:emod
27 | 49 umls:ortf
28 | 1 _CHOICE
29 | 18 umls:enzy
30 | 41 tmtool:ProteinMutation
31 | 24 umls:horm
32 | 47 umls:diap
33 | 10 umls:orch
34 | 16 umls:bact
35 | 33 tmtool:Gene
36 | 56 umls:cell
37 | 39 umls:qlco
38 | 8 umls:phsu
39 | 14 _QUANTITY
40 | 31 umls:phsf
41 | 54 umls:inpr
42 | 37 umls:rcpt
43 | 6 umls:tisu
44 | 12 umls:neop
45 | 29 umls:topp
46 | 52 umls:bsoj
47 | 35 umls:medd
48 | 58 umls:mbrt
49 | 4 umls:bacs
50 | 21 umls:euka
51 | 44 umls:elii
52 | 27 umls:mobd
53 | 50 umls:amas
54 | 2 umls:nnon
55 | 19 umls:qnco
56 | 42 umls:dora
57 | 25 umls:imft
58 | 48 umls:geoa
59 | 11 umls:celc
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/answer_type/4b-dev-liblinear.lindex:
--------------------------------------------------------------------------------
1 | 23 umls:gngm
2 | 46 umls:geoa
3 | 69 umls:resd
4 | 15 umls:elii
5 | 38 umls:fish
6 | 61 tmtool:Disease
7 | 84 umls:inpo
8 | 7 umls:phsu
9 | 30 umls:patf
10 | 53 umls:inbe
11 | 76 umls:mbrt
12 | 22 umls:blor
13 | 45 umls:lbpr
14 | 34 umls:antb
15 | 57 umls:inpr
16 | 80 umls:anab
17 | 3 umls:bacs
18 | 26 umls:dsyn
19 | 49 umls:ortf
20 | 72 umls:prog
21 | 18 _QUANTITY
22 | 41 umls:comd
23 | 64 umls:resa
24 | 10 umls:celc
25 | 33 umls:rcpt
26 | 56 umls:amas
27 | 79 umls:phpr
28 | 68 umls:nusq
29 | 14 umls:diap
30 | 37 umls:cgab
31 | 60 tmtool:Gene
32 | 83 tmtool:FamilyName
33 | 6 umls:aapp
34 | 29 umls:ftcn
35 | 52 umls:clna
36 | 75 umls:orga
37 | 21 umls:qlco
38 | 44 umls:imft
39 | 67 umls:hlca
40 | 2 umls:nnon
41 | 25 umls:amph
42 | 48 umls:phsf
43 | 71 tmtool:Species
44 | 17 umls:medd
45 | 40 umls:menp
46 | 63 umls:bsoj
47 | 9 umls:orch
48 | 32 umls:bact
49 | 55 umls:irda
50 | 78 umls:bmod
51 | 13 umls:fngs
52 | 36 umls:moft
53 | 59 umls:inch
54 | 82 umls:tmco
55 | 5 umls:tisu
56 | 28 umls:mobd
57 | 51 umls:orgf
58 | 74 tmtool:SNP
59 | 20 umls:vita
60 | 43 umls:sosy
61 | 66 umls:dora
62 | 1 _CHOICE
63 | 24 umls:food
64 | 47 umls:emod
65 | 70 tmtool:ProteinMutation
66 | 16 umls:enzy
67 | 39 umls:sbst
68 | 62 umls:bdsu
69 | 85 umls:npop
70 | 8 umls:bpoc
71 | 31 umls:euka
72 | 54 umls:genf
73 | 77 umls:idcn
74 | 12 umls:topp
75 | 35 umls:qnco
76 | 58 tmtool:Chemical
77 | 81 umls:plnt
78 | 4 umls:fndg
79 | 27 umls:clas
80 | 50 umls:celf
81 | 73 umls:cell
82 | 19 umls:virs
83 | 42 umls:horm
84 | 65 umls:spco
85 | 11 umls:neop
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/concept/rerank/4b-dev-liblinear.lindex:
--------------------------------------------------------------------------------
1 | 2 false
2 | 1 true
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/concept/rerank/4b-dev-weka-logistic.dataset-schema:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/concept/rerank/4b-dev-weka-logistic.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/concept/rerank/4b-dev-weka-logistic.model:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/concept/rerank/4b-dev-weka-logistic.model
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/document/rerank/4b-dev-liblinear.lindex:
--------------------------------------------------------------------------------
1 | 2 true
2 | 1 false
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/document/rerank/4b-dev-weka-logistic.dataset-schema:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/document/rerank/4b-dev-weka-logistic.dataset-schema
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/document/rerank/4b-dev-weka-logistic.model:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/oaqa/bioasq/e77fbbb8b9e85226e31f53b16dd2a79a77d59f1b/src/main/resources/models/bioqa/document/rerank/4b-dev-weka-logistic.model
--------------------------------------------------------------------------------
/src/main/resources/models/bioqa/passage/rerank/4b-dev-liblinear.lindex:
--------------------------------------------------------------------------------
1 | 2 false
2 | 1 true
--------------------------------------------------------------------------------
/src/main/resources/properties/direct-gopubmed-concept.properties:
--------------------------------------------------------------------------------
1 | #
2 | # Open Advancement Question Answering (OAQA) Project Copyright 2016 Carnegie Mellon University
3 | #
4 | # Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
5 | # in compliance with the License. You may obtain a copy of the License at
6 | #
7 | # http://www.apache.org/licenses/LICENSE-2.0
8 | #
9 | # Unless required by applicable law or agreed to in writing, software distributed under the License
10 | # is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
11 | # or implied. See the License for the specific language governing permissions and limitations
12 | # under the License.
13 | #
14 |
15 | disease.ontology.service = http://bioasq.org:8000/do
16 | gene.ontology.service = http://bioasq.org:8000/go
17 | jochem.service = http://bioasq.org:8000/jochem
18 | mesh.service = http://bioasq.org:8000/mesh
19 | uniprot.service = http://bioasq.org:8000/uniprot
20 | require.session.url = true
21 | session.refresh.interval = 9
22 |
--------------------------------------------------------------------------------
/src/main/resources/properties/direct-gopubmed-concept.properties.old:
--------------------------------------------------------------------------------
1 | disease.ontology.service = http://gopubmed.org/web/bioasq/doid/json
2 | gene.ontology.service = http://gopubmed.org/web/bioasq/go/json
3 | jochem.service = http://gopubmed.org/web/bioasq/jochem/json
4 | mesh.service = http://gopubmed.org/web/bioasq/mesh/json
5 | uniprot.service = http://gopubmed.org/web/bioasq/uniprot/json
6 | require.session.url = true
7 | session.refresh.interval = 9
8 |
--------------------------------------------------------------------------------
/src/main/resources/properties/direct-gopubmed-document.properties:
--------------------------------------------------------------------------------
1 | #
2 | # Open Advancement Question Answering (OAQA) Project Copyright 2016 Carnegie Mellon University
3 | #
4 | # Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
5 | # in compliance with the License. You may obtain a copy of the License at
6 | #
7 | # http://www.apache.org/licenses/LICENSE-2.0
8 | #
9 | # Unless required by applicable law or agreed to in writing, software distributed under the License
10 | # is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
11 | # or implied. See the License for the specific language governing permissions and limitations
12 | # under the License.
13 | #
14 |
15 | pubmed.search.service = http://bioasq.org:8000/pubmed
16 | require.session.url = true
17 | session.refresh.interval = 9
18 |
--------------------------------------------------------------------------------
/src/main/resources/properties/direct-gopubmed-document.properties.old:
--------------------------------------------------------------------------------
1 | pubmed.search.service = http://gopubmed.org/web/gopubmedbeta/bioasq/pubmed
2 | require.session.url = true
3 | session.refresh.interval = 9
4 |
--------------------------------------------------------------------------------
/src/main/resources/properties/direct-gopubmed-triple.properties:
--------------------------------------------------------------------------------
1 | #
2 | # Open Advancement Question Answering (OAQA) Project Copyright 2016 Carnegie Mellon University
3 | #
4 | # Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except
5 | # in compliance with the License. You may obtain a copy of the License at
6 | #
7 | # http://www.apache.org/licenses/LICENSE-2.0
8 | #
9 | # Unless required by applicable law or agreed to in writing, software distributed under the License
10 | # is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
11 | # or implied. See the License for the specific language governing permissions and limitations
12 | # under the License.
13 | #
14 |
15 | linked.life.data.service = http://gopubmed.org/web/bioasq/linkedlifedata2/triples
16 | require.session.url = true
17 | session.refresh.interval = 9
18 |
--------------------------------------------------------------------------------