├── .gitignore ├── LICENSE ├── README.md ├── doc ├── allclasses-frame.html ├── allclasses-noframe.html ├── constant-values.html ├── deprecated-list.html ├── eu │ └── newsreader │ │ └── eventcoreference │ │ ├── coref │ │ ├── ChainCorefSets.html │ │ ├── ComponentMatch.html │ │ ├── DomainMatch.html │ │ ├── GranularityMatch.html │ │ ├── NormalizeScores.html │ │ ├── PruneSets.html │ │ ├── package-frame.html │ │ ├── package-summary.html │ │ └── package-tree.html │ │ ├── evaluation │ │ ├── EvaluateCorefs.html │ │ ├── package-frame.html │ │ ├── package-summary.html │ │ └── package-tree.html │ │ ├── input │ │ ├── CorefSaxParser.html │ │ ├── RdfReader.html │ │ ├── SemDomParser.html │ │ ├── SesameTrigReader.html │ │ ├── TrigReader.html │ │ ├── package-frame.html │ │ ├── package-summary.html │ │ └── package-tree.html │ │ ├── naf │ │ ├── ClusterEventObjects.html │ │ ├── EntityCorefReferenceBaseline.html │ │ ├── EventCorefLemmaBaseline.html │ │ ├── GetSemFromNafFile.html │ │ ├── GetSemFromNafStream.html │ │ ├── InterDocumentEntityCoref.html │ │ ├── InterDocumentEventCoref.html │ │ ├── MatchEventObjects.html │ │ ├── ResourcesUri.html │ │ ├── package-frame.html │ │ ├── package-summary.html │ │ └── package-tree.html │ │ ├── objects │ │ ├── AgataMention.html │ │ ├── CoRefSetAgata.html │ │ ├── CompositeEvent.html │ │ ├── CorefTargetAgata.html │ │ ├── NafMention.html │ │ ├── OwlTime.html │ │ ├── PhraseCount.html │ │ ├── Sem.html │ │ ├── SemActor.html │ │ ├── SemEvent.html │ │ ├── SemObject.html │ │ ├── SemPlace.html │ │ ├── SemRelation.html │ │ ├── SemTime.html │ │ ├── SourceMeta.html │ │ ├── Triple.html │ │ ├── package-frame.html │ │ ├── package-summary.html │ │ └── package-tree.html │ │ ├── ontology │ │ ├── NodeData.html │ │ ├── OwlReader.html │ │ ├── RdfSense.html │ │ ├── Tree.html │ │ ├── TreeNode.html │ │ ├── package-frame.html │ │ ├── package-summary.html │ │ └── package-tree.html │ │ ├── output │ │ ├── CombineCorefOutputPerEvent.html │ │ ├── CorefSetToSem.html │ │ ├── CorefTargetsInKafSentences.html │ │ ├── CoreferenceBaseline.html │ │ ├── DocumentToCorpusCoreferences.html │ │ ├── FixSynsetReferences.html │ │ ├── JenaSerialization.html │ │ ├── MergeDocumentToCorpusCoreferencesByLemma.html │ │ ├── package-frame.html │ │ ├── package-summary.html │ │ └── package-tree.html │ │ ├── pwn │ │ ├── PwnSaxParser.html │ │ ├── package-frame.html │ │ ├── package-summary.html │ │ └── package-tree.html │ │ └── util │ │ ├── Anaphor.html │ │ ├── EntityTypes.html │ │ ├── EventTypes.html │ │ ├── FilterCorefSets.html │ │ ├── LabelCount.html │ │ ├── ProcessNewsReaderCarBatch.html │ │ ├── ProcessNewsReaderFolderBatch.html │ │ ├── ProcessNewsReaderTechCrunchBatch.html │ │ ├── ProcessSeparateNafFilesBatch.html │ │ ├── ReadSourceMetaFile.html │ │ ├── RoleLabels.html │ │ ├── SemUtilObject.html │ │ ├── SplitTechCrunch.html │ │ ├── Statistics.TotalStats.html │ │ ├── Statistics.html │ │ ├── TimeLanguage.html │ │ ├── Util.AppendableObjectOutputStream.html │ │ ├── Util.html │ │ ├── package-frame.html │ │ ├── package-summary.html │ │ └── package-tree.html ├── help-doc.html ├── index-files │ ├── index-1.html │ ├── index-10.html │ ├── index-11.html │ ├── index-12.html │ ├── index-13.html │ ├── index-14.html │ ├── index-15.html │ ├── index-16.html │ ├── index-17.html │ ├── index-18.html │ ├── index-19.html │ ├── index-2.html │ ├── index-20.html │ ├── index-3.html │ ├── index-4.html │ ├── index-5.html │ ├── index-6.html │ ├── index-7.html │ ├── index-8.html │ └── index-9.html ├── index.html ├── overview-frame.html ├── overview-summary.html ├── overview-tree.html ├── package-list ├── resources │ └── inherit.gif ├── serialized-form.html └── stylesheet.css ├── install.sh ├── pom.xml ├── scripts ├── event-coreference-en.sh ├── event-coreference-lemma.sh ├── event-coreference-nl.sh ├── event-coreference-singleton.sh ├── jena-log4j.properties ├── naf-folder-to-simple-trig.sh ├── naf-folder-to-trig.sh ├── naf2sem-batch-cluster.sh ├── naf2sem-batch-nocluster.sh ├── naf2sem-grasp.sh ├── naf2sem-usage.sh ├── run_naf_folder.sh ├── run_phase1.sh ├── run_single_naf.sh └── trigstatistics.sh ├── src └── main │ └── java │ ├── META-INF │ └── MANIFEST.MF │ └── eu │ └── newsreader │ └── eventcoreference │ ├── coref │ ├── ChainCorefSets.java │ ├── ComponentMatch.java │ ├── Scoring.java │ └── WithinDocumentEventCoref.java │ ├── evaluation │ └── EvaluateCorefs.java │ ├── input │ ├── CorefSaxParser.java │ ├── EsoReader.java │ ├── FrameNetReader.java │ ├── GetEntityCounts.java │ ├── GetEntityStats.java │ ├── GetEsoCounts.java │ ├── GetEventStats.java │ ├── GetNAF.java │ ├── NarrativeChains.java │ ├── OwlReader.java │ ├── OwlReader2.java │ ├── RdfReader.java │ ├── SelectEntityTrig.java │ ├── SemDomParser.java │ ├── SesameTrigReader.java │ ├── TrigKSTripleReader.java │ ├── TrigStatsReader.java │ ├── TrigTripleData.java │ ├── TrigTripleReader.java │ └── TrigUtil.java │ ├── naf │ ├── ClusterEventObjects.java │ ├── CreateMicrostory.java │ ├── EntityCorefReferenceBaseline.java │ ├── EventCorefLemmaBaseline.java │ ├── EventCorefSingletonBaseline.java │ ├── EventCorefWordnetSim.java │ ├── GetJsonTimeLinesFromNaf.java │ ├── GetPerspectiveRelations.java │ ├── GetPerspectiveStatsFromNafFile.java │ ├── GetSemFromCat.java │ ├── GetSemFromNaf.java │ ├── GetSemFromNafFile.java │ ├── GetSemFromNafFolder.java │ ├── GetSemFromNafStream.java │ ├── GetSimpleSemFromNafFolder.java │ ├── GetSimpleSemFromNafStream.java │ ├── GetSimpleSemFromOldBaileyNafFolder.java │ ├── GetTimeLinesFromNaf.java │ ├── GetTimeLinesFromNafFolder.java │ ├── MatchEventObjects.java │ ├── NafSemParameters.java │ ├── NoClusterEventObjects.java │ ├── ParcAttributionTask.java │ ├── ProcessEventObjectsStream.java │ ├── ResourcesUri.java │ └── StructuredEventObjects.java │ ├── objects │ ├── AgataMention.java │ ├── CoRefSetAgata.java │ ├── CompositeEvent.java │ ├── CorefMatch.java │ ├── CorefResultSet.java │ ├── CorefTargetAgata.java │ ├── CorefTriple.java │ ├── JsonEvent.java │ ├── NafMention.java │ ├── OwlTime.java │ ├── PerspectiveObject.java │ ├── PhraseCount.java │ ├── Sem.java │ ├── SemActor.java │ ├── SemEvent.java │ ├── SemObject.java │ ├── SemPlace.java │ ├── SemRelation.java │ ├── SemTime.java │ ├── SourceMeta.java │ ├── TimeTypes.java │ └── Triple.java │ ├── ontology │ └── thesis │ │ ├── NodeData.java │ │ ├── OwlReader.java │ │ ├── RdfSense.java │ │ ├── Tree.java │ │ └── TreeNode.java │ ├── output │ ├── CombineCorefOutputPerEvent.java │ ├── CorefSetToSem.java │ ├── CorefTargetsInKafSentences.java │ ├── CoreferenceBaseline.java │ ├── DataSetConceptHierarchy.java │ ├── DataSetEntityHierarchy.java │ ├── DataSetEventHierarchy.java │ ├── DataSetSources.java │ ├── DataSetTopicHierarchy.java │ ├── DocumentToCorpusCoreferences.java │ ├── FixSynsetReferences.java │ ├── GetEventCentricGraph.java │ ├── GetTypeHierarchyForConcepts.java │ ├── JenaSerialization.java │ ├── MergeDocumentToCorpusCoreferencesByLemma.java │ ├── SimpleTaxonomy.java │ ├── TrigStats.java │ └── WriteStatementsKnowledgeStore.java │ ├── pwn │ ├── ILIReader.java │ └── PwnSaxParser.java │ ├── referencenet │ ├── ReferenceAnnotation.java │ ├── ReferenceNet.java │ └── Util.java │ ├── storyline │ ├── JsonFromCat.java │ ├── JsonFromRdf.java │ ├── JsonSerialization.java │ ├── JsonStoryUtil.java │ ├── MentionResolver.java │ ├── NafTokenLayerIndex.java │ ├── PerspectiveJsonObject.java │ ├── QueryKnowledgeStoreToJsonStoryPerspectives.java │ ├── QueryKnowledgeStoreToJsonStoryPerspectivesOld.java │ ├── ReadFtData.java │ ├── TrigToJsonStoryPerspectives.java │ └── TrigToJsonTimeLine.java │ ├── timeline │ └── EntityTimeLine.java │ └── util │ ├── Anaphor.java │ ├── Clusterer.java │ ├── CompareTrigFiles.java │ ├── EntityTypes.java │ ├── EuroVoc.java │ ├── EventTypes.java │ ├── FilterCorefSets.java │ ├── FixUri.java │ ├── FrameTypes.java │ ├── LabelCount.java │ ├── MD5Checksum.java │ ├── MapEuroVocToEuroConcept.java │ ├── PipelineOverview.java │ ├── ProcessNewsReaderFolderBatch.java │ ├── ProcessSeparateNafFilesBatch.java │ ├── ReadSourceMetaFile.java │ ├── RemoveEventCoreferences.java │ ├── RoleLabels.java │ ├── SemUtilObject.java │ ├── SplitTechCrunch.java │ ├── Statistics.java │ ├── TimeLanguage.java │ ├── TopicStatsFromTrig.java │ ├── TreeStaticHtml.java │ └── Util.java └── usage /.gitignore: -------------------------------------------------------------------------------- 1 | *.class 2 | .idea/ 3 | /eventcoreference.properties 4 | /EventCoreference.ipr 5 | /eventcoreference.xml 6 | /.gitignore 7 | /EventCoreference.iws 8 | /EventCoreference.iml 9 | EventCoreference.iws 10 | EventCoreference.ipr 11 | eventcoreference.properties 12 | eventcoreference.xml 13 | module_eventcoreference.xml 14 | target/ 15 | *.iml 16 | doc/EventCorefWordnetSimTest.html 17 | doc/eu/newsreader/eventcoreference/coref/Scoring.html 18 | doc/eu/newsreader/eventcoreference/input/EsoReader.html 19 | doc/eu/newsreader/eventcoreference/input/FrameNetReader.html 20 | doc/eu/newsreader/eventcoreference/input/GetEntityCounts.html 21 | doc/eu/newsreader/eventcoreference/input/GetEntityStats.html 22 | doc/eu/newsreader/eventcoreference/input/GetEsoCounts.html 23 | doc/eu/newsreader/eventcoreference/input/GetEventStats.html 24 | doc/eu/newsreader/eventcoreference/input/GetNAF.html 25 | doc/eu/newsreader/eventcoreference/input/NarrativeChains.html 26 | doc/eu/newsreader/eventcoreference/input/OwlReader.html 27 | doc/eu/newsreader/eventcoreference/input/OwlReader2.html 28 | doc/eu/newsreader/eventcoreference/input/SelectEntityTrig.html 29 | doc/eu/newsreader/eventcoreference/input/TrigKSTripleReader.html 30 | doc/eu/newsreader/eventcoreference/input/TrigStatsReader.html 31 | doc/eu/newsreader/eventcoreference/input/TrigTripleData.html 32 | doc/eu/newsreader/eventcoreference/input/TrigTripleReader.html 33 | doc/eu/newsreader/eventcoreference/input/TrigUtil.html 34 | doc/eu/newsreader/eventcoreference/naf/CreateMicrostory.html 35 | doc/eu/newsreader/eventcoreference/naf/EventCorefSingletonBaseline.html 36 | doc/eu/newsreader/eventcoreference/naf/EventCorefWordnetSim.html 37 | doc/eu/newsreader/eventcoreference/naf/GetJsonTimeLinesFromNaf.html 38 | doc/eu/newsreader/eventcoreference/naf/GetPerspectiveFromNafFile.html 39 | doc/eu/newsreader/eventcoreference/naf/GetPerspectiveRelations.html 40 | doc/eu/newsreader/eventcoreference/naf/GetPerspectiveStatsFromNafFile.PerspectiveStats.html 41 | doc/eu/newsreader/eventcoreference/naf/GetPerspectiveStatsFromNafFile.html 42 | doc/eu/newsreader/eventcoreference/naf/GetSemAndPerspectiveFromNafFile.html 43 | doc/eu/newsreader/eventcoreference/naf/GetSemFromNaf.html 44 | doc/eu/newsreader/eventcoreference/naf/GetSemFromNafFolder.html 45 | doc/eu/newsreader/eventcoreference/naf/GetTimeLinesFromNaf.html 46 | doc/eu/newsreader/eventcoreference/naf/GetTimeLinesFromNafFolder.html 47 | doc/eu/newsreader/eventcoreference/naf/Naf2SemTest.html 48 | doc/eu/newsreader/eventcoreference/naf/NafObjectToSemTest.html 49 | doc/eu/newsreader/eventcoreference/naf/NoClusterEventObjects.html 50 | doc/eu/newsreader/eventcoreference/naf/ParcAttributionTask.html 51 | doc/eu/newsreader/eventcoreference/naf/ProcessEventObjectsStream.html 52 | doc/eu/newsreader/eventcoreference/naf/RawTextIndex.html 53 | doc/eu/newsreader/eventcoreference/objects/CorefMatch.html 54 | doc/eu/newsreader/eventcoreference/objects/CorefResultSet.html 55 | doc/eu/newsreader/eventcoreference/objects/CorefTriple.html 56 | doc/eu/newsreader/eventcoreference/objects/JsonEvent.html 57 | doc/eu/newsreader/eventcoreference/objects/PerspectiveObject.html 58 | doc/eu/newsreader/eventcoreference/objects/PhraseCount.Compare.html 59 | doc/eu/newsreader/eventcoreference/objects/TimeTypes.html 60 | doc/eu/newsreader/eventcoreference/ontology/thesis/ 61 | doc/eu/newsreader/eventcoreference/output/DataSetConceptHierarchy.html 62 | doc/eu/newsreader/eventcoreference/output/DataSetEntityHierarchy.html 63 | doc/eu/newsreader/eventcoreference/output/DataSetEventHierarchy.html 64 | doc/eu/newsreader/eventcoreference/output/DataSetSources.html 65 | doc/eu/newsreader/eventcoreference/output/DataSetTopicHierarchy.html 66 | doc/eu/newsreader/eventcoreference/output/GetEventCentricGraph.html 67 | doc/eu/newsreader/eventcoreference/output/GetTypeHierarchyForConcepts.html 68 | doc/eu/newsreader/eventcoreference/output/SimpleTaxonomy.html 69 | doc/eu/newsreader/eventcoreference/output/TrigStats.html 70 | doc/eu/newsreader/eventcoreference/pwn/ILIReader.html 71 | doc/eu/newsreader/eventcoreference/storyline/ 72 | doc/eu/newsreader/eventcoreference/timeline/ 73 | doc/eu/newsreader/eventcoreference/util/Clusterer.html 74 | doc/eu/newsreader/eventcoreference/util/CompareTrigFiles.html 75 | doc/eu/newsreader/eventcoreference/util/EuroVoc.html 76 | doc/eu/newsreader/eventcoreference/util/FixUri.html 77 | doc/eu/newsreader/eventcoreference/util/FrameTypes.html 78 | doc/eu/newsreader/eventcoreference/util/MapEuroVocToEuroConcept.html 79 | doc/eu/newsreader/eventcoreference/util/RemoveEventCoreferences.html 80 | doc/eu/newsreader/eventcoreference/util/TopicStatsFromTrig.html 81 | doc/eu/newsreader/eventcoreference/util/TreeStaticHtml.html 82 | doc/index-files/index-21.html 83 | doc/index-files/index-22.html 84 | doc/index-files/index-23.html 85 | doc/index-files/index-24.html 86 | doc/package-frame.html 87 | doc/package-summary.html 88 | doc/package-tree.html 89 | doc/script.js 90 | lib/ 91 | src/main/java/eu/newsreader/eventcoreference/evaluation/.svn/ 92 | src/main/java/eu/newsreader/eventcoreference/input/.svn/ 93 | src/main/java/eu/newsreader/eventcoreference/objects/.svn/ 94 | src/main/java/eu/newsreader/eventcoreference/output/.svn/ 95 | src/main/java/eu/newsreader/eventcoreference/pwn/.svn/ 96 | src/main/java/eu/newsreader/eventcoreference/util/.svn/ 97 | src/main/resources/ 98 | -------------------------------------------------------------------------------- /doc/eu/newsreader/eventcoreference/coref/package-frame.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | eu.newsreader.eventcoreference.coref 7 | 8 | 9 | 10 | 11 | 12 |

eu.newsreader.eventcoreference.coref

13 |
14 |

Classes

15 | 20 |
21 | 22 | 23 | -------------------------------------------------------------------------------- /doc/eu/newsreader/eventcoreference/coref/package-tree.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | eu.newsreader.eventcoreference.coref Class Hierarchy 7 | 8 | 9 | 10 | 11 | 12 | 22 | 25 | 26 |
27 | 28 | 29 |
Skip navigation links
30 | 31 | 32 | 33 | 42 |
43 |
44 | 48 | 52 | 55 |
56 | 66 |
67 | 68 | 69 |
70 | 71 |
72 |

Hierarchy For Package eu.newsreader.eventcoreference.coref

73 | Package Hierarchies: 74 | 77 |
78 |
79 |

Class Hierarchy

80 | 89 |
90 | 91 |
92 | 93 | 94 |
Skip navigation links
95 | 96 | 97 | 98 | 107 |
108 |
109 | 113 | 117 | 120 |
121 | 131 |
132 | 133 | 134 |
135 | 136 | 137 | 138 | -------------------------------------------------------------------------------- /doc/eu/newsreader/eventcoreference/evaluation/package-frame.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | eu.newsreader.eventcoreference.evaluation 7 | 8 | 9 | 10 | 11 | 12 |

eu.newsreader.eventcoreference.evaluation

13 |
14 |

Classes

15 | 18 |
19 | 20 | 21 | -------------------------------------------------------------------------------- /doc/eu/newsreader/eventcoreference/evaluation/package-summary.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | eu.newsreader.eventcoreference.evaluation 7 | 8 | 9 | 10 | 11 | 12 | 22 | 25 | 26 |
27 | 28 | 29 |
Skip navigation links
30 | 31 | 32 | 33 | 42 |
43 |
44 | 48 | 52 | 55 |
56 | 66 |
67 | 68 | 69 |
70 | 71 |
72 |

Package eu.newsreader.eventcoreference.evaluation

73 |
74 |
75 | 94 |
95 | 96 |
97 | 98 | 99 |
Skip navigation links
100 | 101 | 102 | 103 | 112 |
113 |
114 | 118 | 122 | 125 |
126 | 136 |
137 | 138 | 139 |
140 | 141 | 142 | 143 | -------------------------------------------------------------------------------- /doc/eu/newsreader/eventcoreference/evaluation/package-tree.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | eu.newsreader.eventcoreference.evaluation Class Hierarchy 7 | 8 | 9 | 10 | 11 | 12 | 22 | 25 | 26 |
27 | 28 | 29 |
Skip navigation links
30 | 31 | 32 | 33 | 42 |
43 |
44 | 48 | 52 | 55 |
56 | 66 |
67 | 68 | 69 |
70 | 71 |
72 |

Hierarchy For Package eu.newsreader.eventcoreference.evaluation

73 | Package Hierarchies: 74 | 77 |
78 |
79 |

Class Hierarchy

80 | 87 |
88 | 89 |
90 | 91 | 92 |
Skip navigation links
93 | 94 | 95 | 96 | 105 |
106 |
107 | 111 | 115 | 118 |
119 | 129 |
130 | 131 | 132 |
133 | 134 | 135 | 136 | -------------------------------------------------------------------------------- /doc/eu/newsreader/eventcoreference/input/package-frame.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | eu.newsreader.eventcoreference.input 7 | 8 | 9 | 10 | 11 | 12 |

eu.newsreader.eventcoreference.input

13 |
14 |

Classes

15 | 37 |
38 | 39 | 40 | -------------------------------------------------------------------------------- /doc/eu/newsreader/eventcoreference/naf/package-frame.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | eu.newsreader.eventcoreference.naf 7 | 8 | 9 | 10 | 11 | 12 |

eu.newsreader.eventcoreference.naf

13 |
14 |

Classes

15 | 40 |
41 | 42 | 43 | -------------------------------------------------------------------------------- /doc/eu/newsreader/eventcoreference/objects/package-frame.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | eu.newsreader.eventcoreference.objects 7 | 8 | 9 | 10 | 11 | 12 |

eu.newsreader.eventcoreference.objects

13 |
14 |

Classes

15 | 40 |
41 | 42 | 43 | -------------------------------------------------------------------------------- /doc/eu/newsreader/eventcoreference/ontology/package-frame.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | eu.newsreader.eventcoreference.ontology 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | eu.newsreader.eventcoreference.ontology 20 | 21 | 22 | 35 | 36 |
23 | Classes  24 | 25 |
26 | NodeData 27 |
28 | OwlReader 29 |
30 | RdfSense 31 |
32 | Tree 33 |
34 | TreeNode
37 | 38 | 39 | 40 | 41 | -------------------------------------------------------------------------------- /doc/eu/newsreader/eventcoreference/output/package-frame.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | eu.newsreader.eventcoreference.output 7 | 8 | 9 | 10 | 11 | 12 |

eu.newsreader.eventcoreference.output

13 |
14 |

Classes

15 | 34 |
35 | 36 | 37 | -------------------------------------------------------------------------------- /doc/eu/newsreader/eventcoreference/pwn/package-frame.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | eu.newsreader.eventcoreference.pwn 7 | 8 | 9 | 10 | 11 | 12 |

eu.newsreader.eventcoreference.pwn

13 |
14 |

Classes

15 | 19 |
20 | 21 | 22 | -------------------------------------------------------------------------------- /doc/eu/newsreader/eventcoreference/util/package-frame.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | eu.newsreader.eventcoreference.util 7 | 8 | 9 | 10 | 11 | 12 |

eu.newsreader.eventcoreference.util

13 |
14 |

Classes

15 | 41 |
42 | 43 | 44 | -------------------------------------------------------------------------------- /doc/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | Generated Documentation (Untitled) 7 | 59 | 60 | 61 | 62 | 63 | 64 | 65 | 66 | 67 | <noscript> 68 | <div>JavaScript is disabled on your browser.</div> 69 | </noscript> 70 | <h2>Frame Alert</h2> 71 | <p>This document is designed to be viewed using the frames feature. If you see this message, you are using a non-frame-capable web client. Link to <a href="overview-summary.html">Non-frame version</a>.</p> 72 | 73 | 74 | 75 | -------------------------------------------------------------------------------- /doc/overview-frame.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | Overview List 7 | 8 | 9 | 10 | 11 | 12 |
All Classes
13 |
14 |

Packages

15 | 28 |
29 |

 

30 | 31 | 32 | -------------------------------------------------------------------------------- /doc/package-list: -------------------------------------------------------------------------------- 1 | eu.newsreader.eventcoreference.coref 2 | eu.newsreader.eventcoreference.evaluation 3 | eu.newsreader.eventcoreference.input 4 | eu.newsreader.eventcoreference.naf 5 | eu.newsreader.eventcoreference.objects 6 | eu.newsreader.eventcoreference.ontology.thesis 7 | eu.newsreader.eventcoreference.output 8 | eu.newsreader.eventcoreference.pwn 9 | eu.newsreader.eventcoreference.storyline 10 | eu.newsreader.eventcoreference.timeline 11 | eu.newsreader.eventcoreference.util 12 | -------------------------------------------------------------------------------- /doc/resources/inherit.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/cltl/EventCoreference/f1de443975dfde8c017b12ee4e93b3ad34ecb777/doc/resources/inherit.gif -------------------------------------------------------------------------------- /doc/stylesheet.css: -------------------------------------------------------------------------------- 1 | /* Javadoc style sheet */ 2 | 3 | /* Define colors, fonts and other style attributes here to override the defaults */ 4 | 5 | /* Page background color */ 6 | body { background-color: #FFFFFF; color:#000000 } 7 | 8 | /* Headings */ 9 | h1 { font-size: 145% } 10 | 11 | /* Table colors */ 12 | .TableHeadingColor { background: #CCCCFF; color:#000000 } /* Dark mauve */ 13 | .TableSubHeadingColor { background: #EEEEFF; color:#000000 } /* Light mauve */ 14 | .TableRowColor { background: #FFFFFF; color:#000000 } /* White */ 15 | 16 | /* Font used in left-hand frame lists */ 17 | .FrameTitleFont { font-size: 100%; font-family: Helvetica, Arial, sans-serif; color:#000000 } 18 | .FrameHeadingFont { font-size: 90%; font-family: Helvetica, Arial, sans-serif; color:#000000 } 19 | .FrameItemFont { font-size: 90%; font-family: Helvetica, Arial, sans-serif; color:#000000 } 20 | 21 | /* Navigation bar fonts and colors */ 22 | .NavBarCell1 { background-color:#EEEEFF; color:#000000} /* Light mauve */ 23 | .NavBarCell1Rev { background-color:#00008B; color:#FFFFFF} /* Dark Blue */ 24 | .NavBarFont1 { font-family: Arial, Helvetica, sans-serif; color:#000000;color:#000000;} 25 | .NavBarFont1Rev { font-family: Arial, Helvetica, sans-serif; color:#FFFFFF;color:#FFFFFF;} 26 | 27 | .NavBarCell2 { font-family: Arial, Helvetica, sans-serif; background-color:#FFFFFF; color:#000000} 28 | .NavBarCell3 { font-family: Arial, Helvetica, sans-serif; background-color:#FFFFFF; color:#000000} 29 | 30 | -------------------------------------------------------------------------------- /install.sh: -------------------------------------------------------------------------------- 1 | set -e 2 | 3 | #Requires an installation of maven 2.x and Java 1.6 or higher 4 | 5 | # define the location of the install scipt 6 | DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )" 7 | PARENT="$( cd $DIR && cd .. && pwd)" 8 | 9 | mkdir -p "$DIR/lib" 10 | echo "#1. compiling the library from source code and dependencies" 11 | mvn clean 12 | mvn install 13 | echo "#2. moving binary to lib folder" 14 | jarfile=`find $DIR/target -name "EventCoreference-v*-jar-with-dependencies.jar" -print` 15 | if 16 | [ "$jarfile" == "" ] 17 | then 18 | echo "could not generate the jar file. Sorry ..." 19 | exit 4 20 | fi 21 | mv $jarfile $DIR/lib 22 | echo "#3. installing the vua-resources" 23 | cd "$PARENT" 24 | if 25 | [ -d "vua-resources" ] 26 | then 27 | cd vua-resources 28 | git pull 29 | cd .. 30 | else 31 | git clone https://github.com/cltl/vua-resources.git 32 | fi 33 | echo "#4. cleaning up" 34 | cd "$DIR" 35 | rm -R target 36 | rm -R src 37 | rm EventCoreference.iml 38 | rm EventCoreference.iws 39 | rm EventCoreference.ipr 40 | -------------------------------------------------------------------------------- /scripts/event-coreference-en.sh: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env bash 2 | DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )" 3 | ROOT="$( cd $DIR && cd .. && pwd)" 4 | LIB="$ROOT"/lib 5 | REL="has_hyperonym#event#has_hypernym" 6 | 7 | RESOURCES="$( cd $ROOT && cd .. && pwd)"/vua-resources 8 | # assumes vua-resources is installed next to this installation 9 | # git clone https://github.com/cltl/vua-resources.git 10 | 11 | #pass naf file as input stream and catch the naf output stream 12 | # for example> "cat example-naf.xml | event-coreference.sh > example-naf.coref.xml" 13 | 14 | java -Xmx812m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.EventCorefWordnetSim --method leacock-chodorow --wn-lmf "$RESOURCES/wneng-30.lmf.xml.xpos.gz" --sim 2.0 --sim-ont 0.6 --wsd 0.8 --relations $REL --wn-prefix "eng" --source-frames "$RESOURCES/source.txt" --grammatical-frames "$RESOURCES/grammatical.txt" --contextual-frames "$RESOURCES/contextual.txt" 15 | -------------------------------------------------------------------------------- /scripts/event-coreference-lemma.sh: -------------------------------------------------------------------------------- 1 | DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )" 2 | ROOT="$( cd $DIR && cd .. && pwd)" 3 | LIB="$ROOT"/lib 4 | 5 | #pass naf file as input stream and catch the naf output stream 6 | # for example> "cat example-naf.xml | event-coreference-lemma.sh > example-naf.coref.xml" 7 | 8 | java -Xmx812m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.EventCorefLemmaBaseline 9 | 10 | -------------------------------------------------------------------------------- /scripts/event-coreference-nl.sh: -------------------------------------------------------------------------------- 1 | DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )" 2 | ROOT="$( cd $DIR && cd .. && pwd)" 3 | LIB="$ROOT"/lib 4 | REL="XPOS_NEAR_SYNONYM#HAS_HYPERONYM#HAS_XPOS_HYPERONYM#event" 5 | 6 | RESOURCES="$( cd $ROOT && cd .. && pwd)"/vua-resources 7 | # assumes vua-resources is installed next to this installation 8 | # git clone https://github.com/cltl/vua-resources.git 9 | 10 | #pass naf file as input stream and catch the naf output stream 11 | # for example> "cat example-naf.xml | event-coreference.sh > example-naf.coref.xml" 12 | 13 | java -Xmx812m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.EventCorefWordnetSim --method leacock-chodorow --wn-lmf "$RESOURCES/odwn_orbn_gwg-LMF_1.3.xml.gz" --sim 2.0 --sim-ont 0.6 --wsd 0.8 --relations $REL --source-frames "$RESOURCES/source.txt" --grammatical-frames "$RESOURCES/grammatical.txt" --contextual-frames "$RESOURCES/contextual.txt" 14 | -------------------------------------------------------------------------------- /scripts/event-coreference-singleton.sh: -------------------------------------------------------------------------------- 1 | DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )" 2 | ROOT="$( cd $DIR && cd .. && pwd)" 3 | LIB="$ROOT"/lib 4 | 5 | #pass naf file as input stream and catch the naf output stream 6 | # for example> "cat example-naf.xml | event-coreference-lemma.sh > example-naf.coref.xml" 7 | 8 | java -Xmx812m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.EventCorefSingletonBaseline 9 | 10 | -------------------------------------------------------------------------------- /scripts/jena-log4j.properties: -------------------------------------------------------------------------------- 1 | log4j.rootLogger=INFO, stdlog 2 | ## log4j.rootLogger=INFO, FusekiFileLog 3 | 4 | log4j.appender.stdlog=org.apache.log4j.ConsoleAppender 5 | ## log4j.appender.stdlog.target=System.err 6 | log4j.appender.stdlog.layout=org.apache.log4j.PatternLayout 7 | log4j.appender.stdlog.layout.ConversionPattern=%d{HH:mm:ss} %-5p %-20c{1} :: %m%n 8 | 9 | ## # Example for file logging. 10 | ## log4j.appender.FusekiFileLog=org.apache.log4j.DailyRollingFileAppender 11 | ## log4j.appender.FusekiFileLog.DatePattern='.'yyyy-MM-dd 12 | ## log4j.appender.FusekiFileLog.File=fuseki-log 13 | ## log4j.appender.FusekiFileLog.layout=org.apache.log4j.PatternLayout 14 | ## log4j.appender.FusekiFileLog.layout.ConversionPattern=%d{HH:mm:ss} %-5p %-20c{1} :: %m%n 15 | 16 | # Jetty - Fuseki catches Jetty errors and reports them. 17 | log4j.logger.org.eclipse.jetty=FATAL 18 | 19 | # Execution logging 20 | log4j.logger.com.hp.hpl.jena.arq.info=INFO 21 | log4j.logger.com.hp.hpl.jena.arq.exec=INFO 22 | 23 | # TDB loader 24 | log4j.logger.com.hp.hpl.jena.tdb.loader=INFO 25 | # TDB syslog. 26 | log4j.logger.TDB=INFO 27 | 28 | # Everything else in Jena 29 | log4j.logger.com.hp.hpl.jena=WARN 30 | log4j.logger.org.openjena=WARN 31 | log4j.logger.org.openjena.riot=INFO 32 | 33 | # Fuseki 34 | # Server log. 35 | log4j.logger.org.openjena.fuseki.Server=INFO 36 | # Request log. 37 | log4j.logger.org.openjena.fuseki.Fuseki=INFO 38 | # Internal logs 39 | log4j.logger.org.openjena.fuseki=INFO 40 | -------------------------------------------------------------------------------- /scripts/naf-folder-to-simple-trig.sh: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env bash 2 | DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )" 3 | ROOT="$( cd $DIR && cd .. && pwd)" 4 | LIB="$ROOT"/lib 5 | 6 | RESOURCES="$( cd $ROOT && cd .. && pwd)"/vua-resources 7 | # assumes vua-resources is installed next to this installation 8 | # git clone https://github.com/cltl/vua-resources.git 9 | # 10 | # the software assume that the input files are in NAF format minially with the following layers: 11 | # - tokens, terms, entities, coreference for events, srl, timeExpressions 12 | # to create an event coreference layer, use the event-coreference scripts 13 | 14 | #DUTCH 15 | ontology ="$RESOURCES"/dbpedia_nl_types.tsv.gz 16 | #ENGLISH 17 | ontology ="$RESOURCES"/instance_types_en.ttl.gz 18 | java -Xmx2000m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.GetSimpleSemFromNafFolder --naf-folder $1 --extension ".naf" --ontology $ontology 19 | -------------------------------------------------------------------------------- /scripts/naf-folder-to-trig.sh: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env bash 2 | DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )" 3 | ROOT="$( cd $DIR && cd .. && pwd)" 4 | LIB="$ROOT"/lib 5 | 6 | RESOURCES="$( cd $ROOT && cd .. && pwd)"/vua-resources 7 | # assumes vua-resources is installed next to this installation 8 | # git clone https://github.com/cltl/vua-resources.git 9 | # 10 | # the software assume that the input files are in NAF format minially with the following layers: 11 | # - tokens, terms, entities, coreference for events, srl, timeExpressions 12 | # to create an event coreference layer, use the event-coreference scripts 13 | 14 | java -Xmx2000m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.GetSemFromNafFolder --naf-folder $1 --extension ".naf" --non-entities --project wikinews --all --ili "$RESOURCES/ili.ttl.gz" --perspective --eurovoc-en "$RESOURCES/mapping_eurovoc_skos.label.concept.gz" --source-frames "$RESOURCES/source.txt" 15 | -------------------------------------------------------------------------------- /scripts/naf2sem-batch-cluster.sh: -------------------------------------------------------------------------------- 1 | DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )" 2 | ROOT="$( cd $DIR && cd .. && pwd)" 3 | LIB="$ROOT"/lib 4 | EVENTS="$1" 5 | PROJECT="$2" 6 | RESOURCES="$( cd $ROOT && cd .. && pwd)"/vua-resources 7 | # assumes vua-resources is installed next to this installation 8 | # git clone https://github.com/cltl/vua-resources.git 9 | # the software assume that the input files are in NAF format minially with the following layers: 10 | # - tokens, terms, entities, coreference for events, srl, timeExpressions 11 | # to create an event coreference layer, use the event-coreference scripts 12 | 13 | 14 | java -Xmx2000m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.ClusterEventObjects --naf-folder "$EVENTS" --event-folder "$EVENTS" --extension ".xml" --non-entities --project "$PROJECT" --all --eurovoc-en "$RESOURCES/mapping_eurovoc_skos.csv.gz" --source-frames "$RESOURCES/source.txt" --grammatical-frames "$RESOURCES/grammatical.txt" --contextual-frames "$RESOURCES/contextual.txt" 15 | 16 | java -Xmx2000m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.MatchEventObjects --event-folder "$EVENTS/events/contextualEvent" --match-type ililemma --roles "anyrole" --concept-match 30 --phrase-match 30 --hypers --lcs --ili "$RESOURCES//ili.ttl.gz" 17 | 18 | java -Xmx2000m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.MatchEventObjects --event-folder "$EVENTS/events/sourceEvent" --match-type ililemma --roles "a0" --concept-match 30 --phrase-match 30 --ili "$RESOURCES//ili.ttl.gz" 19 | 20 | java -Xmx2000m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.MatchEventObjects --event-folder "$EVENTS/events/grammaticalEvent" --match-type lemma --roles "a1" --phrase-match 30 --ili "$RESOURCES/ili.ttl.gz" 21 | 22 | java -Xmx2000m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.MatchEventObjects --event-folder "$EVENTS/events/futureEvent" --match-type lemma --roles "all" --phrase-match 30 --ili "$RESOURCES/ili.ttl.gz" 23 | -------------------------------------------------------------------------------- /scripts/naf2sem-batch-nocluster.sh: -------------------------------------------------------------------------------- 1 | DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )" 2 | ROOT="$( cd $DIR && cd .. && pwd)" 3 | LIB="$ROOT"/lib 4 | EVENTS="$1" 5 | PROJECT="$2" 6 | RESOURCES="$( cd $ROOT && cd .. && pwd)"/vua-resources 7 | # assumes vua-resources is installed next to this installation 8 | # git clone https://github.com/cltl/vua-resources.git 9 | # the software assume that the input files are in NAF format minially with the following layers: 10 | # - tokens, terms, entities, coreference for events, srl, timeExpressions 11 | # to create an event coreference layer, use the event-coreference scripts 12 | 13 | java -Xmx2000m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.NoClusterEventObjects --naf-folder "$EVENTS" --event-folder "$EVENTS" --extension ".xml" --non-entities --project "$PROJECT" --all --eurovoc-en "$RESOURCES/mapping_eurovoc_skos.csv.gz" 14 | 15 | java -Xmx2000m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.MatchEventObjects --event-folder "$EVENTS" --match-type ililemma --roles "anyrole" --concept-match 30 --phrase-match 30 --ili "$RESOURCES/ili.ttl.gz" --hypers --lcs 16 | 17 | -------------------------------------------------------------------------------- /scripts/naf2sem-grasp.sh: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env bash 2 | DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )" 3 | ROOT="$( cd $DIR && cd .. && pwd)" 4 | LIB="$ROOT"/lib 5 | 6 | RESOURCES="$( cd $ROOT && cd .. && pwd)"/vua-resources 7 | # assumes vua-resources is installed next to this installation 8 | # git clone https://github.com/cltl/vua-resources.git 9 | # 10 | # the software assume that the input files are in NAF format minially with the following layers: 11 | # - tokens, terms, entities, coreference for events, srl, timeExpressions 12 | # to create an event coreference layer, use the event-coreference scripts 13 | 14 | java -Xmx2000m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.GetSemFromNafStream --non-entities --project test --all --ili "$RESOURCES/ili.ttl.gz" --perspective --eurovoc-en "$RESOURCES/mapping_eurovoc_skos.label.concept.gz" --source-frames "$RESOURCES/source.txt" 15 | -------------------------------------------------------------------------------- /scripts/naf2sem-usage.sh: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env bash 2 | DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )" 3 | ROOT="$( cd $DIR && cd .. && pwd)" 4 | LIB="$ROOT"/lib 5 | 6 | 7 | 8 | java -Xmx2000m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.naf.NafSemParameters 9 | -------------------------------------------------------------------------------- /scripts/run_naf_folder.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | 3 | extensions=(naf) 4 | 5 | for ext in "${extensions[@]}" 6 | do 7 | echo "Look for files with extension: $ext" 8 | if [ -e ../nafs_to_process/*.$ext ]; then 9 | echo "Files with extension $ext exist. Now processing them ..." 10 | for f in ../nafs_to_process/*.$ext ; 11 | do 12 | base=$(basename $f) 13 | echo "File: $base" 14 | cat $f | java -Xmx2000m -cp ../target/EventCoreference-v3.1.2-jar-with-dependencies.jar eu.newsreader.eventcoreference.naf.GetSemFromNafStream --project cars --source-frames "../resources/source.txt" --grammatical-frames "../resources/grammatical.txt" --contextual-frames "../resources/contextual.txt" --non-entities --timex-max 5 --ili ../resources/ili.ttl | java -Xmx2000m -cp ../target/EventCoreference-1.0-SNAPSHOT-jar-with-dependencies.jar eu.newsreader.eventcoreference.naf.ProcessEventObjectsStream --source-roles "pb\:A0,pb\:A1" --contextual-match-type "LEMMA" > "../trigs/$base.trig" 15 | echo "Trig created. Now inserting into the KS" 16 | wget -O /dev/null --post-file "../trigs/$base.trig" --header 'Content-type: application/x-trig' https://knowledgestore2.fbk.eu/nwr/aitor/custom/naf2sem 17 | 18 | if [ $? -ne 0 ]; then 19 | echo "Error occurred while inserting into the KS. Please review the naf and the trig files..." 20 | else 21 | echo "Insertion done" 22 | mv $f ../nafs_processed/ 23 | fi 24 | 25 | done 26 | else 27 | echo "No files with extension $ext exist in the directory ../nafs_to_process! " 28 | fi 29 | done 30 | -------------------------------------------------------------------------------- /scripts/run_phase1.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | 3 | extensions=(naf) 4 | 5 | for ext in "${extensions[@]}" 6 | do 7 | echo "Look for files with extension: $ext" 8 | if [ -e ../nafs_to_process/*.$ext ]; then 9 | echo "Files with extension $ext exist. Now processing them ..." 10 | for f in ../nafs_to_process/*.$ext ; 11 | do 12 | base=$(basename $f) 13 | echo "File: $base" 14 | cat $f | java -Xmx2000m -cp ../target/EventCoreference-v3.1.2-jar-with-dependencies.jar eu.newsreader.eventcoreference.naf.GetSemFromNafStream --project cars --source-frames "../resources/source.txt" --grammatical-frames "../resources/grammatical.txt" --contextual-frames "../resources/contextual.txt" --non-entities --timex-max 5 --ili ../resources/ili.ttl > ../phase1.trig 15 | echo "Trig created. Now inserting into the KS" 16 | #wget -O /dev/null --post-file "../trigs/$base.trig" --header 'Content-type: application/x-trig' https://knowledgestore2.fbk.eu/nwr/aitor/custom/naf2sem 17 | #echo "Insertion done" 18 | # mv $f ../nafs_processed/ 19 | done 20 | else 21 | echo "No files with extension $ext exist in the directory ../nafs_to_process! " 22 | fi 23 | done 24 | -------------------------------------------------------------------------------- /scripts/run_single_naf.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | 3 | f="$1" 4 | 5 | if [ -z $1 ]; then 6 | echo "Please specify a file. Exiting now..." 7 | exit 8 | fi 9 | 10 | base=$(basename $f) 11 | echo "File to process: $base" 12 | cat $f | java -Xmx2000m -cp ../target/EventCoreference-v3.1.2-jar-with-dependencies.jar eu.newsreader.eventcoreference.naf.GetSemFromNafStream --project end-to-end --source-frames "../resources/source.txt" --grammatical-frames "../resources/grammatical.txt" --contextual-frames "../resources/contextual.txt" --non-entities --perspective --ili ../resources/ili.ttl.gz --eurovoc-en "../resources/mapping_eurovoc_skos.csv" | java -Xmx2000m -cp ../target/EventCoreference-3.0-jar-with-dependencies.jar eu.newsreader.eventcoreference.naf.ProcessEventObjectsStream --source-roles "pb\:A0,pb\:A1" --contextual-match-type "ILILEMMA" --ks https://myknowledgestore --user me --passw mypassw > "../trigs/$base.trig" 13 | 14 | echo "Trig created. Now inserting into the KS" 15 | wget -O /dev/null --post-file "../trigs/$base.trig" --header 'Content-type: application/x-trig' https://knowledgestore2.fbk.eu/nwr/aitor/custom/naf2sem 16 | if [ $? -ne 0 ]; then 17 | echo "Error occurred while inserting into the KS. Please review the naf and the trig files..." 18 | else 19 | echo "Insertion done" 20 | mv $f ../nafs_processed/ 21 | fi 22 | -------------------------------------------------------------------------------- /scripts/trigstatistics.sh: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env bash 2 | DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )" 3 | ROOT="$( cd $DIR && cd .. && pwd)" 4 | LIB="$ROOT"/target 5 | 6 | RESOURCES="$( cd $ROOT && cd .. && pwd)"/vua-resources 7 | 8 | PATH_TO_TRIG=$1 9 | TYPE="instance" 10 | # instance or grasp 11 | 12 | #STAT= 13 | # can be empty (dumps all) or "en=entities", "ne=non-entities", "dbp=dbpedia" , "event" or "topic" 14 | 15 | # limit the freqquency 16 | N="0" 17 | 18 | java -Xmx2000m -cp "$LIB/EventCoreference-v3.1.2-jar-with-dependencies.jar" eu.newsreader.eventcoreference.output.TrigStats --trig-folder $PATH_TO_TRIG --type $TYPE --n $N 19 | #--stat $STAT 20 | 21 | -------------------------------------------------------------------------------- /src/main/java/META-INF/MANIFEST.MF: -------------------------------------------------------------------------------- 1 | Manifest-Version: 1.0 2 | Main-Class: eu.newsreader.eventcoreference.naf.GetSemFromNafStream 3 | 4 | -------------------------------------------------------------------------------- /src/main/java/eu/newsreader/eventcoreference/coref/ChainCorefSets.java: -------------------------------------------------------------------------------- 1 | package eu.newsreader.eventcoreference.coref; 2 | 3 | import eu.newsreader.eventcoreference.objects.CorefMatch; 4 | import eu.newsreader.eventcoreference.objects.CorefResultSet; 5 | 6 | import java.util.ArrayList; 7 | 8 | /** 9 | * Created with IntelliJ IDEA. 10 | * User: kyoto 11 | * Date: 2/17/13 12 | * Time: 6:12 PM 13 | * To change this template use File | Settings | File Templates. 14 | */ 15 | public class ChainCorefSets { 16 | 17 | static public ArrayList chainResultSets (ArrayList corefMatchList) { 18 | ArrayList condensedSets = new ArrayList(); 19 | for (int i = 0; i < corefMatchList.size(); i++) { 20 | CorefResultSet candidateCorefResultSet = corefMatchList.get(i); 21 | boolean CHAINIT = false; 22 | for (int j = 0; j < condensedSets.size(); j++) { 23 | CorefResultSet coreferenceSet = condensedSets.get(j); 24 | /// do we need to do something with the sources??????? 25 | /* if (coreferenceSet.hasSpanAndLowestCommonSubsumer(corefMatch)) { 26 | //if (coreferenceSet.hasSpan(corefMatch.getCorefTargets())) { 27 | //// there is a match so start adding the corefResultSet to the KafCoreferenceSet 28 | CHAINIT = true; 29 | break; 30 | } 31 | else { 32 | //if span matches but not the lcs we need to choose one of the other 33 | }*/ 34 | for (int k = 0; k < candidateCorefResultSet.getTargets().size(); k++) { 35 | CorefMatch corefMatch = candidateCorefResultSet.getTargets().get(k); 36 | if (coreferenceSet.hasSpanAndLowestCommonSubsumer(corefMatch)) { 37 | //if (coreferenceSet.hasSpan(corefMatch.getCorefTargets())) { 38 | //// there is a match so start adding the corefResultSet to the KafCoreferenceSet 39 | CHAINIT = true; 40 | break; 41 | } 42 | else { 43 | //if span matches but not the lcs we need to choose one of the other 44 | } 45 | } 46 | if (CHAINIT) { 47 | for (int k = 0; k < candidateCorefResultSet.getTargets().size(); k++) { 48 | CorefMatch corefMatch = candidateCorefResultSet.getTargets().get(k); 49 | coreferenceSet.addTargetScore(corefMatch); 50 | } 51 | // coreferenceSet.addSourceCoref(candidateCorefResultSet.getSources()); 52 | /// we are done with this set and can break the loop 53 | break; 54 | } 55 | } 56 | if (!CHAINIT) { 57 | condensedSets.add(candidateCorefResultSet); 58 | } 59 | } 60 | return condensedSets; 61 | } 62 | 63 | 64 | 65 | } 66 | -------------------------------------------------------------------------------- /src/main/java/eu/newsreader/eventcoreference/coref/Scoring.java: -------------------------------------------------------------------------------- 1 | package eu.newsreader.eventcoreference.coref; 2 | 3 | import eu.newsreader.eventcoreference.objects.CorefMatch; 4 | 5 | import java.util.ArrayList; 6 | 7 | /** 8 | * Created by piek on 10/13/14. 9 | */ 10 | public class Scoring { 11 | static public ArrayList pruneCoref (ArrayList corefsets, int threshold) { 12 | ArrayList newCorefMatches = new ArrayList(); 13 | for (int i = 0; i < corefsets.size(); i++) { 14 | CorefMatch corefMatch = corefsets.get(i); 15 | if (corefMatch.getScore()*100>=threshold) { 16 | newCorefMatches.add(corefMatch); 17 | } 18 | else { 19 | // System.out.println("removing corefMatch = " + corefMatch.getScore()); 20 | } 21 | } 22 | return newCorefMatches; 23 | } 24 | 25 | static public void normalizeCorefMatch (ArrayList coRefSets) { 26 | double maxScore = 0; 27 | double minScore = 0; 28 | for (int c = 0; c < coRefSets.size(); c++) { 29 | CorefMatch corefMatch = coRefSets.get(c); 30 | double corefScore = corefMatch.getScore(); 31 | if (corefScore>maxScore) { 32 | maxScore = corefScore; 33 | } 34 | else if (corefScore(i+1)) { 33 | user = args[i+1]; 34 | } 35 | else if (arg.equalsIgnoreCase("--p") && args.length>(i+1)) { 36 | pass = args[i+1]; 37 | } 38 | } 39 | Authenticator.setDefault(new Authenticator() { 40 | 41 | @Override 42 | protected PasswordAuthentication getPasswordAuthentication() { 43 | return new PasswordAuthentication(user, pass.toCharArray()); 44 | } 45 | }); 46 | 47 | 48 | System.out.println(getFile("https://knowledgestore2.fbk.eu/nwr/cars3/files?id=%3Chttp%3A%2F%2Fwww.newsreader-project.eu%2Fdata%2Fcars%2F2004%2F10%2F18%2F4DKT-30W0-00S0-W39B.xml.naf%3E")); 49 | long estimatedTime = System.currentTimeMillis() - startTime; 50 | System.out.println("Time elapsed:"); 51 | System.out.println(estimatedTime/1000.0); 52 | } 53 | 54 | public static String getFile(String stringUrl) throws Exception { 55 | //stringUrl = "https://knowledgestore2.fbk.eu/nwr/cars3/files?id=%3Chttp%3A%2F%2Fwww.newsreader-project.eu%2Fdata%2Fcars%2F2004%2F10%2F18%2F4DKT-30W0-00S0-W39B.xml.naf%3E"; 56 | 57 | URL url = new URL(stringUrl); 58 | HttpURLConnection connection = 59 | (HttpURLConnection) url.openConnection(); 60 | connection.setRequestMethod("GET"); 61 | // connection.setRequestProperty("Accept", "application/xml"); 62 | connection.setRequestProperty("Accept", "application/octet-stream"); 63 | InputStream xml = connection.getInputStream(); 64 | KafSaxParser kafSaxParser = new KafSaxParser(); 65 | kafSaxParser.parseFile(xml); 66 | String rawText = kafSaxParser.rawText; 67 | /* DocumentBuilderFactory dbf = DocumentBuilderFactory.newInstance(); 68 | DocumentBuilder db = dbf.newDocumentBuilder(); 69 | Document doc = db.parse(xml); 70 | return getStringFromDocument(doc);*/ 71 | return rawText; 72 | } 73 | 74 | public static ArrayList getNafWordFormsFile(String stringUrl) throws Exception { 75 | //stringUrl = "https://knowledgestore2.fbk.eu/nwr/cars3/files?id=%3Chttp%3A%2F%2Fwww.newsreader-project.eu%2Fdata%2Fcars%2F2004%2F10%2F18%2F4DKT-30W0-00S0-W39B.xml.naf%3E"; 76 | KafSaxParser kafSaxParser = new KafSaxParser(); 77 | 78 | URL url = new URL(stringUrl); 79 | HttpURLConnection connection = 80 | (HttpURLConnection) url.openConnection(); 81 | connection.setRequestMethod("GET"); 82 | // connection.setRequestProperty("Accept", "application/xml"); 83 | /* connection.setRequestProperty("Accept", "application/octet-stream"); 84 | InputStream xml = connection.getInputStream(); 85 | kafSaxParser.parseFile(xml);*/ 86 | 87 | connection.setRequestProperty("Accept-Encoding", "gzip"); /// gets gzipped NAF 88 | InputStream xml = connection.getInputStream(); 89 | InputStream gzipStream = new GZIPInputStream(xml); 90 | kafSaxParser.parseFile(gzipStream); 91 | System.out.println(gzipStream); 92 | return kafSaxParser.kafWordFormList; 93 | } 94 | 95 | public static String getText(String stringUrl) throws Exception { 96 | //stringUrl = "https://knowledgestore2.fbk.eu/nwr/cars3/files?id=%3Chttp%3A%2F%2Fwww.newsreader-project.eu%2Fdata%2Fcars%2F2004%2F10%2F18%2F4DKT-30W0-00S0-W39B.xml.naf%3E"; 97 | 98 | URL url = new URL(stringUrl); 99 | HttpURLConnection connection = 100 | (HttpURLConnection) url.openConnection(); 101 | connection.setRequestMethod("GET"); 102 | // connection.setRequestProperty("Accept", "application/xml"); 103 | connection.setRequestProperty("Accept", "application/octet-stream"); 104 | InputStream xml = connection.getInputStream(); 105 | String rawText = xml.toString(); 106 | return rawText; 107 | } 108 | 109 | 110 | public static String getStringFromDocument(Document doc) throws TransformerException { 111 | DOMSource domSource = new DOMSource(doc); 112 | StringWriter writer = new StringWriter(); 113 | StreamResult result = new StreamResult(writer); 114 | TransformerFactory tf = TransformerFactory.newInstance(); 115 | Transformer transformer = tf.newTransformer(); 116 | transformer.transform(domSource, result); 117 | return writer.toString(); 118 | } 119 | 120 | } 121 | -------------------------------------------------------------------------------- /src/main/java/eu/newsreader/eventcoreference/input/NarrativeChains.java: -------------------------------------------------------------------------------- 1 | package eu.newsreader.eventcoreference.input; 2 | 3 | import java.io.*; 4 | import java.util.ArrayList; 5 | import java.util.HashMap; 6 | 7 | /** 8 | * Created by piek on 09/05/15. 9 | */ 10 | public class NarrativeChains { 11 | //Events: sell buy own acquire operate purchase spin_off build plan pay merge announce 12 | 13 | public HashMap> chainMap; 14 | public HashMap> verbMap; 15 | 16 | public NarrativeChains (String chainFilePath) { 17 | chainMap = new HashMap>(); 18 | verbMap = new HashMap>(); 19 | if (new File(chainFilePath).exists() ) { 20 | try { 21 | FileInputStream fis = new FileInputStream(chainFilePath); 22 | InputStreamReader isr = new InputStreamReader(fis); 23 | BufferedReader in = new BufferedReader(isr); 24 | String inputLine; 25 | Integer chainId = 0; 26 | while (in.ready()&&(inputLine = in.readLine()) != null) { 27 | //System.out.println(inputLine); 28 | if (inputLine.trim().startsWith("Events:")) { 29 | chainId++; 30 | ArrayList verbsArrayList = new ArrayList(); 31 | String [] verbs = inputLine.split(" "); 32 | for (int i = 1; i < verbs.length; i++) { /// skipping Events: 33 | String verb = verbs[i]; 34 | verbsArrayList.add(verb); 35 | if (chainMap.containsKey(verb)) { 36 | ArrayList chains = chainMap.get(verb); 37 | chains.add(chainId); 38 | chainMap.put(verb, chains); 39 | } 40 | else { 41 | ArrayList chains = new ArrayList(); 42 | chains.add(chainId); 43 | chainMap.put(verb, chains); 44 | } 45 | } 46 | verbMap.put(chainId, verbsArrayList); 47 | } 48 | } 49 | in.close(); 50 | } catch (IOException e) { 51 | e.printStackTrace(); 52 | } 53 | } 54 | } 55 | 56 | 57 | } 58 | -------------------------------------------------------------------------------- /src/main/java/eu/newsreader/eventcoreference/input/OwlReader.java: -------------------------------------------------------------------------------- 1 | package eu.newsreader.eventcoreference.input; 2 | 3 | import com.hp.hpl.jena.ontology.OntClass; 4 | import com.hp.hpl.jena.ontology.OntModel; 5 | import com.hp.hpl.jena.ontology.OntModelSpec; 6 | import com.hp.hpl.jena.query.QueryExecution; 7 | import com.hp.hpl.jena.query.QueryExecutionFactory; 8 | import com.hp.hpl.jena.query.QuerySolution; 9 | import com.hp.hpl.jena.query.ResultSet; 10 | import com.hp.hpl.jena.rdf.model.ModelFactory; 11 | import com.hp.hpl.jena.rdf.model.Resource; 12 | import com.hp.hpl.jena.util.iterator.Filter; 13 | import eu.newsreader.eventcoreference.naf.ResourcesUri; 14 | 15 | import java.util.Iterator; 16 | 17 | /** 18 | * Created by piek on 10/5/14. 19 | */ 20 | public class OwlReader { 21 | 22 | 23 | static public String createSparql(String variable, String typeUri) { 24 | String q = "SELECT ?"+variable+" " + 25 | "WHERE{" + 26 | "?"+variable+" a "+typeUri + 27 | "}"; 28 | return q; 29 | } 30 | 31 | static public String createSparqlTriple(String variable, String subj, String pred, String obj) { 32 | String q = "SELECT ?"+variable+" \n" + 33 | "WHERE { " + 34 | " "+subj+ " \n"+ 35 | " "+pred+ " \n"+ 36 | " "+obj+ 37 | "}"; 38 | return q; 39 | } 40 | 41 | static void readOwlFile (String pathToOwlFile) { 42 | OntModel ontologyModel = 43 | ModelFactory.createOntologyModel(OntModelSpec.OWL_MEM, null); 44 | ontologyModel.read(pathToOwlFile, "RDF/XML-ABBREV"); 45 | // OntClass myClass = ontologyModel.getOntClass("namespace+className"); 46 | 47 | OntClass myClass = ontologyModel.getOntClass(ResourcesUri.nwr+"domain-ontology#Motion"); 48 | System.out.println("myClass.toString() = " + myClass.toString()); 49 | System.out.println("myClass.getSuperClass().toString() = " + myClass.getSuperClass().toString()); 50 | 51 | //List list = 52 | // namedHierarchyRoots(ontologyModel); 53 | 54 | 55 | Iterator i = ontologyModel.listHierarchyRootClasses() 56 | .filterDrop( new Filter() { 57 | public boolean accept( Object o ) { 58 | return ((Resource) o).isAnon(); 59 | }} ); ///get all top nodes and excludes anonymous classes 60 | 61 | // Iterator i = ontologyModel.listHierarchyRootClasses(); 62 | while (i.hasNext()) { 63 | System.out.println(i.next().toString()); 64 | /* OntClass ontClass = ontologyModel.getOntClass(i.next().toString()); 65 | if (ontClass.hasSubClass()) { 66 | 67 | }*/ 68 | } 69 | 70 | String q = createSparql("event", ""); 71 | System.out.println("q = " + q); 72 | QueryExecution qe = QueryExecutionFactory.create(q, 73 | ontologyModel); 74 | for (ResultSet rs = qe.execSelect() ; rs.hasNext() ; ) { 75 | QuerySolution binding = rs.nextSolution(); 76 | System.out.println("binding = " + binding.toString()); 77 | System.out.println("Event: " + binding.get("event")); 78 | } 79 | 80 | ontologyModel.close(); 81 | } 82 | 83 | static public void main (String [] args) { 84 | String pathToOwlOntology = "/Users/piek/Desktop/NWR/NWR-ontology/version-0.6/ESO_version_0.6.owl"; 85 | readOwlFile (pathToOwlOntology); 86 | 87 | } 88 | } 89 | -------------------------------------------------------------------------------- /src/main/java/eu/newsreader/eventcoreference/input/OwlReader2.java: -------------------------------------------------------------------------------- 1 | package eu.newsreader.eventcoreference.input; 2 | 3 | 4 | import org.semanticweb.owlapi.apibinding.OWLManager; 5 | import org.semanticweb.owlapi.model.*; 6 | import org.semanticweb.owlapi.reasoner.OWLReasoner; 7 | import org.semanticweb.owlapi.reasoner.OWLReasonerFactory; 8 | 9 | import java.io.PrintStream; 10 | 11 | //import javax.annotation.Nonnull; 12 | 13 | /** 14 | * Created by piek on 10/5/14. 15 | */ 16 | @SuppressWarnings("javadoc") 17 | 18 | 19 | public class OwlReader2 { 20 | 21 | 22 | private static final int INDENT = 4; 23 | // @Nonnull 24 | private final OWLReasonerFactory reasonerFactory; 25 | // @Nonnull 26 | private final OWLOntology ontology; 27 | private final PrintStream out; 28 | 29 | private OwlReader2(OWLReasonerFactory reasonerFactory, 30 | OWLOntology inputOntology) { 31 | this.reasonerFactory = reasonerFactory; 32 | ontology = inputOntology; 33 | out = System.out; 34 | } 35 | 36 | /** 37 | * Print the class hierarchy for the given ontology from this class down, 38 | * assuming this class is at the given level. Makes no attempt to deal 39 | * sensibly with multiple inheritance. 40 | */ 41 | private void printHierarchy(OWLClass clazz) throws OWLException { 42 | OWLReasoner reasoner = reasonerFactory 43 | .createNonBufferingReasoner(ontology); 44 | printHierarchy(reasoner, clazz, 0); 45 | /* Now print out any unsatisfiable classes */ 46 | for (OWLClass cl : ontology.getClassesInSignature()) { 47 | assert cl != null; 48 | if (!reasoner.isSatisfiable(cl)) { 49 | out.println("XXX: " + labelFor(cl)); 50 | } 51 | } 52 | reasoner.dispose(); 53 | } 54 | 55 | private String labelFor(OWLClass clazz) { 56 | /* 57 | * Use a visitor to extract label annotations 58 | */ 59 | /*LabelExtractor le = new LabelExtractor(); 60 | for (OWLAnnotation anno : annotations(ontology 61 | .getAnnotationAssertionAxioms(clazz.getIRI()))) { 62 | anno.accept(le); 63 | } 64 | *//* Print out the label if there is one. If not, just use the class URI *//* 65 | if (le.getResult() != null) { 66 | return le.getResult(); 67 | } else { 68 | return clazz.getIRI().toString(); 69 | }*/ 70 | return ""; 71 | } 72 | 73 | /** 74 | * Print the class hierarchy from this class down, assuming this class is at 75 | * the given level. Makes no attempt to deal sensibly with multiple 76 | * inheritance. 77 | */ 78 | private void printHierarchy(OWLReasoner reasoner, 79 | OWLClass clazz, int level) throws OWLException { 80 | /* 81 | * Only print satisfiable classes -- otherwise we end up with bottom 82 | * everywhere 83 | */ 84 | if (reasoner.isSatisfiable(clazz)) { 85 | for (int i = 0; i < level * INDENT; i++) { 86 | out.print(" "); 87 | } 88 | out.println(labelFor(clazz)); 89 | /* Find the children and recurse */ 90 | for (OWLClass child : reasoner.getSubClasses(clazz, true) 91 | .getFlattened()) { 92 | if (!child.equals(clazz)) { 93 | printHierarchy(reasoner, child, level + 1); 94 | } 95 | } 96 | } 97 | } 98 | 99 | static public void main (String[] args) throws OWLException, 100 | InstantiationException, IllegalAccessException, 101 | ClassNotFoundException { 102 | String reasonerFactoryClassName = null; 103 | 104 | String pathToOwlOntology = "/Users/piek/Desktop/NWR/NWR-ontology/version-0.6/ESO_version_0.6.owl"; 105 | OWLOntologyManager m = (OWLOntologyManager) OWLManager.getOWLDataFactory(); 106 | @SuppressWarnings("null") 107 | //@Nonnull 108 | IRI documentIRI = IRI.create(pathToOwlOntology); 109 | OWLOntology ontology = m 110 | .loadOntologyFromOntologyDocument(documentIRI); 111 | // Report information about the ontology 112 | System.out.println("Ontology Loaded..."); 113 | System.out.println("Document IRI: " + documentIRI); 114 | System.out.println("Ontology : " + ontology.getOntologyID()); 115 | System.out.println("Format : " 116 | + m.getOntologyFormat(ontology)); 117 | 118 | @SuppressWarnings("null") 119 | 120 | OwlReader2 simpleHierarchy = new OwlReader2( 121 | (OWLReasonerFactory) Class.forName(reasonerFactoryClassName) 122 | .newInstance(), ontology); 123 | // Get Thing 124 | OWLClass clazz = m.getOWLDataFactory().getOWLThing(); 125 | System.out.println("Class : " + clazz); 126 | // Print the hierarchy below thing 127 | simpleHierarchy.printHierarchy(clazz); 128 | } 129 | } 130 | -------------------------------------------------------------------------------- /src/main/java/eu/newsreader/eventcoreference/input/RdfReader.java: -------------------------------------------------------------------------------- 1 | package eu.newsreader.eventcoreference.input; 2 | 3 | import com.hp.hpl.jena.rdf.model.Model; 4 | import com.hp.hpl.jena.rdf.model.ModelFactory; 5 | import com.hp.hpl.jena.util.FileManager; 6 | 7 | import java.io.InputStream; 8 | 9 | /** 10 | * Created by piek on 1/3/14. 11 | */ 12 | public class RdfReader { 13 | 14 | 15 | static void readRdfFile (String pathToRdfFile) { 16 | // create an empty model 17 | Model model = ModelFactory.createDefaultModel(); 18 | 19 | // use the FileManager to find the input file 20 | InputStream in = FileManager.get().open( pathToRdfFile ); 21 | if (in == null) { 22 | throw new IllegalArgumentException( 23 | "File: " + pathToRdfFile + " not found"); 24 | } 25 | 26 | // read the RDF/XML file 27 | model.read(in, null); 28 | 29 | // write it to standard out 30 | model.write(System.out); 31 | } 32 | 33 | 34 | static public void main (String[] args) { 35 | // String pathToRdfFile = args[0]; 36 | String pathToRdfFile = "/Users/piek/Desktop/example.rdf"; 37 | readRdfFile(pathToRdfFile); 38 | } 39 | } 40 | -------------------------------------------------------------------------------- /src/main/java/eu/newsreader/eventcoreference/input/SelectEntityTrig.java: -------------------------------------------------------------------------------- 1 | package eu.newsreader.eventcoreference.input; 2 | 3 | import com.hp.hpl.jena.query.Dataset; 4 | import com.hp.hpl.jena.rdf.model.Model; 5 | import com.hp.hpl.jena.rdf.model.Statement; 6 | import com.hp.hpl.jena.rdf.model.StmtIterator; 7 | import com.hp.hpl.jena.tdb.TDBFactory; 8 | import eu.newsreader.eventcoreference.util.Util; 9 | import org.apache.jena.riot.RDFDataMgr; 10 | 11 | import java.io.*; 12 | import java.util.ArrayList; 13 | 14 | /** 15 | * Created by piek on 15/06/15. 16 | */ 17 | public class SelectEntityTrig { 18 | 19 | 20 | static Dataset dataset = null; 21 | 22 | static public void main (String[] args) { 23 | 24 | String trigfolderPath = ""; 25 | trigfolderPath = "/Users/piek/Desktop/tweede-kamer/events"; 26 | String entity = ""; 27 | String date = ""; 28 | entity = "lippens"; 29 | for (int i = 0; i < args.length; i++) { 30 | String arg = args[i]; 31 | if (arg.equals("--trig-folder") && args.length>(i+1)) { 32 | trigfolderPath = args[i+1]; 33 | } 34 | else if (arg.equals("--entity") && args.length>(i+1)) { 35 | entity = args[i+1]; 36 | } 37 | else if (arg.equals("--event-date") && args.length>(i+1)) { 38 | date = args[i+1]; 39 | } 40 | } 41 | if (entity.isEmpty()) { 42 | System.out.println("Entity is empty"); 43 | return; 44 | } 45 | File trigfolder = new File(trigfolderPath); 46 | String trigEntityPath = trigfolder.getParent()+"/"+entity; 47 | File entityTrigFolder = new File (trigEntityPath); 48 | if (!entityTrigFolder.exists()) { 49 | entityTrigFolder.mkdir(); 50 | } 51 | if (!entityTrigFolder.exists()) { 52 | System.out.println("Could not create entity trig folder"); 53 | return; 54 | } 55 | dataset = TDBFactory.createDataset(); 56 | ArrayList trigFiles = Util.makeRecursiveFileList(trigfolder, ".trig"); 57 | System.out.println(trigfolder.getName() + " trigFiles.size() = " + trigFiles.size()); 58 | int cnt = 1; 59 | for (int i = 0; i < trigFiles.size(); i++) { 60 | File file = trigFiles.get(i); 61 | if (!file.getParentFile().getName().startsWith(date)) { 62 | continue; 63 | } 64 | if (i%500==0) { 65 | System.out.println("i = " + i); 66 | // if (i>1000) break; 67 | } 68 | ArrayList events = new ArrayList(); 69 | dataset = RDFDataMgr.loadDataset(file.getAbsolutePath()); 70 | Model namedModel = dataset.getNamedModel(TrigUtil.instanceGraph); 71 | StmtIterator siter = namedModel.listStatements(); 72 | while (siter.hasNext()) { 73 | Statement s = siter.nextStatement(); 74 | String subject = s.getSubject().getURI().toLowerCase(); 75 | if (subject.indexOf(entity.toLowerCase())>-1) { 76 | String trigName = trigEntityPath+"/"+cnt+"_"+file.getName(); 77 | File trigCopy = new File(trigName); 78 | copyFile(file, trigCopy); 79 | cnt++; 80 | break; 81 | } 82 | } 83 | dataset = null; 84 | } 85 | } 86 | 87 | static public void copyFile(File inputFile, File outputFile) { 88 | try { 89 | DataInputStream in = new DataInputStream(new FileInputStream(inputFile)); 90 | byte[] buffer = new byte[(int) inputFile.length()]; 91 | in.readFully(buffer); 92 | in.close(); 93 | DataOutputStream out = new DataOutputStream(new FileOutputStream(outputFile)); 94 | out.write(buffer); 95 | } catch (IOException e) { 96 | e.printStackTrace(); 97 | } 98 | } 99 | 100 | } 101 | -------------------------------------------------------------------------------- /src/main/java/eu/newsreader/eventcoreference/input/TrigTripleData.java: -------------------------------------------------------------------------------- 1 | package eu.newsreader.eventcoreference.input; 2 | 3 | import com.hp.hpl.jena.rdf.model.Statement; 4 | 5 | import java.io.IOException; 6 | import java.io.OutputStream; 7 | import java.util.*; 8 | 9 | /** 10 | * Created by piek on 05/02/16. 11 | */ 12 | public class TrigTripleData { 13 | 14 | 15 | static final public String provenanceGraph = "http://www.newsreader-project.eu/provenance"; 16 | static final public String instanceGraph = "http://www.newsreader-project.eu/instances"; 17 | static final public String graspGraph = "http://www.newsreader-project.eu/grasp"; 18 | public HashMap> tripleMapGrasp = new HashMap>(); 19 | public HashMap> tripleMapProvenance = new HashMap>(); 20 | public HashMap> tripleMapLabels = new HashMap>(); 21 | public HashMap> tripleMapInstances = new HashMap>(); 22 | public HashMap> tripleMapOthers = new HashMap>(); 23 | public Vector perspectiveMentions = new Vector(); 24 | 25 | public TrigTripleData() { 26 | tripleMapLabels = new HashMap>(); 27 | tripleMapGrasp = new HashMap>(); 28 | tripleMapProvenance = new HashMap>(); 29 | tripleMapInstances = new HashMap>(); 30 | tripleMapOthers = new HashMap>(); 31 | perspectiveMentions = new Vector(); 32 | } 33 | 34 | public void dumpTriples (OutputStream fos, HashMap> map) throws IOException { 35 | Set keyset = map.keySet(); 36 | Iterator keys = keyset.iterator(); 37 | while (keys.hasNext()) { 38 | String key = keys.next(); 39 | ArrayList statements = map.get(key); 40 | for (int i = 0; i < statements.size(); i++) { 41 | Statement statement = statements.get(i); 42 | String str = statement.toString()+"\n"; 43 | fos.write(str.getBytes()); 44 | } 45 | } 46 | } 47 | 48 | public void dumpInstanceTriples (OutputStream fos) throws IOException { 49 | Set keyset = tripleMapInstances.keySet(); 50 | Iterator keys = keyset.iterator(); 51 | while (keys.hasNext()) { 52 | String key = keys.next(); 53 | ArrayList statements = tripleMapInstances.get(key); 54 | for (int i = 0; i < statements.size(); i++) { 55 | Statement statement = statements.get(i); 56 | String str = statement.toString()+"\n"; 57 | fos.write(str.getBytes()); 58 | } 59 | } 60 | } 61 | 62 | public void dumpOtherTriples (OutputStream fos) throws IOException { 63 | Set keyset = tripleMapOthers.keySet(); 64 | Iterator keys = keyset.iterator(); 65 | while (keys.hasNext()) { 66 | String key = keys.next(); 67 | ArrayList statements = tripleMapOthers.get(key); 68 | for (int i = 0; i < statements.size(); i++) { 69 | Statement statement = statements.get(i); 70 | String str = statement.toString()+"\n"; 71 | fos.write(str.getBytes()); 72 | } 73 | } 74 | } 75 | } 76 | -------------------------------------------------------------------------------- /src/main/java/eu/newsreader/eventcoreference/naf/GetSemFromNafStream.java: -------------------------------------------------------------------------------- 1 | package eu.newsreader.eventcoreference.naf; 2 | 3 | import eu.kyotoproject.kaf.KafSaxParser; 4 | import eu.newsreader.eventcoreference.coref.ComponentMatch; 5 | import eu.newsreader.eventcoreference.objects.*; 6 | import eu.newsreader.eventcoreference.output.JenaSerialization; 7 | import eu.newsreader.eventcoreference.util.FrameTypes; 8 | import eu.newsreader.eventcoreference.util.MD5Checksum; 9 | import org.apache.jena.atlas.logging.Log; 10 | 11 | import java.util.ArrayList; 12 | 13 | /** 14 | * Created by piek on 2/12/14. 15 | */ 16 | public class GetSemFromNafStream { 17 | 18 | static public void main(String[] args) { 19 | Log.setLog4j("jena-log4j.properties"); 20 | NafSemParameters nafSemParameters = new NafSemParameters(args); 21 | 22 | 23 | ArrayList semEvents = new ArrayList(); 24 | ArrayList semActors = new ArrayList(); 25 | ArrayList semTimes = new ArrayList(); 26 | ArrayList semRelations = new ArrayList(); 27 | KafSaxParser kafSaxParser = new KafSaxParser(); 28 | kafSaxParser.parseFile(System.in); 29 | if (kafSaxParser.getKafMetaData().getUrl().isEmpty()) { 30 | //System.err.println("ERROR! Empty url in header NAF. Cannot create unique URIs! Aborting"); 31 | try { 32 | String checkSum = MD5Checksum.getMD5ChecksumFromString(kafSaxParser.rawText); 33 | // System.err.println("checkSum = " + checkSum); 34 | kafSaxParser.getKafMetaData().setUrl(checkSum); 35 | } catch (Exception e) { 36 | // e.printStackTrace(); 37 | } 38 | } 39 | GetSemFromNaf.processNafFile(nafSemParameters, kafSaxParser, semEvents, semActors, semTimes, semRelations ); 40 | ArrayList compositeEventArraylist = new ArrayList(); 41 | for (int j = 0; j < semEvents.size(); j++) { 42 | SemEvent mySemEvent = (SemEvent) semEvents.get(j); 43 | ArrayList myTimes = ComponentMatch.getMySemTimes(mySemEvent, semRelations, semTimes); 44 | ArrayList myActors = ComponentMatch.getMySemActors(mySemEvent, semRelations, semActors); 45 | ArrayList myRelations = ComponentMatch.getMySemRelations(mySemEvent, semRelations); 46 | CompositeEvent compositeEvent = new CompositeEvent(mySemEvent, myActors, myTimes, myRelations); 47 | if (myTimes.size() <= nafSemParameters.getTIMEEXPRESSIONMAX()) { 48 | if (compositeEvent.isValid()) { 49 | FrameTypes.setEventTypeString(compositeEvent.getEvent(), nafSemParameters); 50 | compositeEventArraylist.add(compositeEvent); 51 | } 52 | else { 53 | /* 54 | System.out.println("Skipping EVENT due to no time anchor and/or no participant"); 55 | System.out.println("compositeEvent = " + compositeEvent.getEvent().getURI()); 56 | System.out.println("myTimes = " + myTimes.size()); 57 | System.out.println("myActors = " + myActors.size()); 58 | System.out.println("myRelations = " + myRelations.size()); 59 | */ 60 | } 61 | } else { 62 | /* 63 | System.out.println("Skipping event due to excessive time expressions linked to it"); 64 | System.out.println("compositeEvent = " + compositeEvent.getEvent().getURI()); 65 | System.out.println("myTimes.size() = " + myTimes.size()); 66 | */ 67 | } 68 | } 69 | if (!nafSemParameters.isPERSPECTIVE()) { 70 | JenaSerialization.serializeJenaCompositeEvents(System.out, compositeEventArraylist, null, nafSemParameters.isILIURI(), nafSemParameters.isVERBOSE()); 71 | } 72 | else { 73 | ArrayList sourcePerspectives = GetPerspectiveRelations.getSourcePerspectives(kafSaxParser, 74 | semActors, 75 | semEvents, 76 | nafSemParameters); 77 | ArrayList documentPerspectives = GetPerspectiveRelations.getAuthorPerspectives( 78 | kafSaxParser, nafSemParameters.getPROJECT(), sourcePerspectives); 79 | JenaSerialization.serializeJenaCompositeEventsAndPerspective(System.out, compositeEventArraylist, kafSaxParser, nafSemParameters.getPROJECT(), sourcePerspectives, documentPerspectives); 80 | } 81 | } 82 | } -------------------------------------------------------------------------------- /src/main/java/eu/newsreader/eventcoreference/naf/GetSimpleSemFromNafStream.java: -------------------------------------------------------------------------------- 1 | package eu.newsreader.eventcoreference.naf; 2 | 3 | import eu.kyotoproject.kaf.KafSaxParser; 4 | import eu.newsreader.eventcoreference.coref.ComponentMatch; 5 | import eu.newsreader.eventcoreference.objects.*; 6 | import eu.newsreader.eventcoreference.output.JenaSerialization; 7 | import eu.newsreader.eventcoreference.output.SimpleTaxonomy; 8 | import eu.newsreader.eventcoreference.util.FrameTypes; 9 | import eu.newsreader.eventcoreference.util.MD5Checksum; 10 | import org.apache.jena.atlas.logging.Log; 11 | 12 | import java.util.ArrayList; 13 | import java.util.HashMap; 14 | 15 | /** 16 | * Created by piek on 2/12/14. 17 | */ 18 | public class GetSimpleSemFromNafStream { 19 | /* 20 | 21 | static final String USAGE = "This program processes a single NAF file and generates SEM RDF-TRiG results" + 22 | "The program has the following arguments:\n" + 23 | "--project \n" + 24 | "--contextual-frames \n" + 25 | "--communication-frames \n" + 26 | "--grammatical-frames " + 27 | ; 28 | 29 | static public Vector sourceVector = null; 30 | static public Vector grammaticalVector = null; 31 | static public Vector contextualVector = null; 32 | static public int TIMEEXPRESSIONMAX = 5; 33 | static public boolean NONENTITIES = false; 34 | static public boolean ILIURI = false; 35 | static public boolean VERBOSE = false; 36 | static public boolean PERSPECTIVE = false; 37 | 38 | static boolean DOCTIME = true; 39 | static boolean CONTEXTTIME = true; 40 | 41 | static boolean NOMCOREF = true; 42 | static boolean EVENTCOREF = true; 43 | */ 44 | static public void main(String[] args) { 45 | Log.setLog4j("jena-log4j.properties"); 46 | NafSemParameters nafSemParameters = new NafSemParameters(args); 47 | 48 | 49 | ArrayList semEvents = new ArrayList(); 50 | ArrayList semActors = new ArrayList(); 51 | ArrayList semTimes = new ArrayList(); 52 | ArrayList semRelations = new ArrayList(); 53 | KafSaxParser kafSaxParser = new KafSaxParser(); 54 | kafSaxParser.parseFile(System.in); 55 | if (kafSaxParser.getKafMetaData().getUrl().isEmpty()) { 56 | //System.err.println("ERROR! Empty url in header NAF. Cannot create unique URIs! Aborting"); 57 | try { 58 | String checkSum = MD5Checksum.getMD5ChecksumFromString(kafSaxParser.rawText); 59 | // System.err.println("checkSum = " + checkSum); 60 | kafSaxParser.getKafMetaData().setUrl(checkSum); 61 | } catch (Exception e) { 62 | // e.printStackTrace(); 63 | } 64 | } 65 | GetSemFromNaf.processNafFile(nafSemParameters, kafSaxParser, semEvents, semActors, semTimes, semRelations ); 66 | ArrayList compositeEventArraylist = new ArrayList(); 67 | for (int j = 0; j < semEvents.size(); j++) { 68 | SemEvent mySemEvent = (SemEvent) semEvents.get(j); 69 | ArrayList myTimes = ComponentMatch.getMySemTimes(mySemEvent, semRelations, semTimes); 70 | ArrayList myActors = ComponentMatch.getMySemActors(mySemEvent, semRelations, semActors); 71 | ArrayList myRelations = ComponentMatch.getMySemRelations(mySemEvent, semRelations); 72 | CompositeEvent compositeEvent = new CompositeEvent(mySemEvent, myActors, myTimes, myRelations); 73 | if (myTimes.size() <= nafSemParameters.getTIMEEXPRESSIONMAX()) { 74 | if (compositeEvent.isValid()) { 75 | FrameTypes.setEventTypeString(compositeEvent.getEvent(), nafSemParameters); 76 | compositeEventArraylist.add(compositeEvent); 77 | } 78 | else { 79 | /* 80 | System.out.println("Skipping EVENT due to no time anchor and/or no participant"); 81 | System.out.println("compositeEvent = " + compositeEvent.getEvent().getURI()); 82 | System.out.println("myTimes = " + myTimes.size()); 83 | System.out.println("myActors = " + myActors.size()); 84 | System.out.println("myRelations = " + myRelations.size()); 85 | */ 86 | } 87 | } else { 88 | /* 89 | System.out.println("Skipping event due to excessive time expressions linked to it"); 90 | System.out.println("compositeEvent = " + compositeEvent.getEvent().getURI()); 91 | System.out.println("myTimes.size() = " + myTimes.size()); 92 | */ 93 | } 94 | } 95 | JenaSerialization.serializeJenaSimpleCompositeEvents(System.out, compositeEventArraylist, new SimpleTaxonomy(), new HashMap()); 96 | } 97 | } -------------------------------------------------------------------------------- /src/main/java/eu/newsreader/eventcoreference/objects/CorefMatch.java: -------------------------------------------------------------------------------- 1 | package eu.newsreader.eventcoreference.objects; 2 | 3 | import eu.kyotoproject.kaf.CorefTarget; 4 | 5 | import java.util.ArrayList; 6 | 7 | /** 8 | * Created by piek on 10/10/14. 9 | */ 10 | public class CorefMatch { 11 | private double score; 12 | private String lowestCommonSubsumer; 13 | private String targetLemma; 14 | private ArrayList> corefTargets; 15 | 16 | public CorefMatch() { 17 | this.score = -1; 18 | this.lowestCommonSubsumer = ""; 19 | this.targetLemma = ""; 20 | this.corefTargets = new ArrayList>(); 21 | } 22 | 23 | public String getTargetLemma() { 24 | return targetLemma; 25 | } 26 | 27 | public void setTargetLemma(String targetLemma) { 28 | this.targetLemma = targetLemma; 29 | } 30 | 31 | public double getScore() { 32 | return score; 33 | } 34 | 35 | public void setScore(double score) { 36 | this.score = score; 37 | } 38 | 39 | public void addScore(double score) { 40 | this.score +=score; 41 | } 42 | 43 | public String getLowestCommonSubsumer() { 44 | return lowestCommonSubsumer; 45 | } 46 | 47 | public void setLowestCommonSubsumer(String lowestCommonSubsumer) { 48 | this.lowestCommonSubsumer = lowestCommonSubsumer; 49 | } 50 | 51 | public ArrayList> getCorefTargets() { 52 | return corefTargets; 53 | } 54 | 55 | public void addCorefTargets(ArrayList corefTargets) { 56 | this.corefTargets.add(corefTargets); 57 | } 58 | 59 | public void setCorefTargets(ArrayList> corefTargets) { 60 | this.corefTargets = corefTargets; 61 | } 62 | 63 | public boolean hasSpan (ArrayList spans) { 64 | for (int i = 0; i < corefTargets.size(); i++) { 65 | ArrayList targets = corefTargets.get(i); 66 | int overlap = CorefTarget.overlapSetOfSpans(targets, spans); 67 | if (overlap==corefTargets.size() && (overlap==spans.size())) { 68 | /// exact match 69 | return true; 70 | } 71 | } 72 | return false; 73 | } 74 | 75 | public boolean hasIntersection (ArrayList spans) { 76 | for (int i = 0; i < corefTargets.size(); i++) { 77 | ArrayList targets = corefTargets.get(i); 78 | int overlap = CorefTarget.overlapSetOfSpans(targets, spans); 79 | if (overlap>0) { 80 | /// exact match 81 | return true; 82 | } 83 | } 84 | return false; 85 | } 86 | } 87 | -------------------------------------------------------------------------------- /src/main/java/eu/newsreader/eventcoreference/objects/CorefTriple.java: -------------------------------------------------------------------------------- 1 | package eu.newsreader.eventcoreference.objects; 2 | 3 | import java.util.ArrayList; 4 | 5 | /** 6 | * Created with IntelliJ IDEA. 7 | * User: kyoto 8 | * Date: 10/11/13 9 | * Time: 6:35 PM 10 | * To change this template use File | Settings | File Templates. 11 | */ 12 | public class CorefTriple { 13 | 14 | private String id; 15 | private String predicate; 16 | private String subject; 17 | private String object; 18 | private ArrayList mentions; 19 | private String comment; 20 | 21 | public CorefTriple() { 22 | this.id = ""; 23 | this.object = ""; 24 | this.predicate = ""; 25 | this.subject = ""; 26 | this.comment = ""; 27 | this.mentions = new ArrayList(); 28 | } 29 | 30 | public CorefTriple(String predicate, String subject, String object) { 31 | this.id = ""; 32 | this.comment = ""; 33 | this.object = object; 34 | this.predicate = predicate; 35 | this.subject = subject; 36 | this.mentions = new ArrayList(); 37 | } 38 | 39 | public String getComment() { 40 | return comment; 41 | } 42 | 43 | public void setComment(String comment) { 44 | this.comment = comment; 45 | } 46 | 47 | public void addComment(String comment) { 48 | this.comment += comment; 49 | } 50 | 51 | public ArrayList getMentions() { 52 | return mentions; 53 | } 54 | 55 | public void setMentions(ArrayList mentions) { 56 | this.mentions = mentions; 57 | } 58 | 59 | public void addMentions(String mention) { 60 | this.mentions.add(mention); 61 | } 62 | 63 | public String getObject() { 64 | return object; 65 | } 66 | 67 | public void setObject(String object) { 68 | this.object = object; 69 | } 70 | 71 | public String getPredicate() { 72 | return predicate; 73 | } 74 | 75 | public void setPredicate(String predicate) { 76 | this.predicate = predicate; 77 | } 78 | 79 | public String getSubject() { 80 | return subject; 81 | } 82 | 83 | public void setSubject(String subject) { 84 | this.subject = subject; 85 | } 86 | 87 | public String getId() { 88 | return id; 89 | } 90 | 91 | public void setId(String id) { 92 | this.id = id; 93 | } 94 | 95 | public String toString () { 96 | String str = "\t\n"; 101 | // str += "