├── README.md └── ReadingNotes.md /README.md: -------------------------------------------------------------------------------- 1 | # awesome-question-answering 2 | 3 | QA 4 | 5 | 领域经典论文，项目及数据集 6 | 7 | #### Papers 8 | - [Memory Networks](http://arxiv.org/pdf/1410.3916v11.pdf) 9 | - [End-To-End Memory Networks](http://arxiv.org/abs/1503.08895) 10 | - [Towards AI-Complete Question Answering: A set of prerequisite toy tasks](http://arxiv.org/pdf/1502.05698v10.pdf) 11 | - [Large Scale simple question answering with Memory Networks](https://arxiv.org/pdf/1506.02075v1.pdf) 12 | - [Ask Me Anything: Dynamic Memory Networks for Natural Language Processing](http://arxiv.org/pdf/1506.07285v5.pdf) 13 | - [Key-Value Memory Networks for directly understanding documents](https://arxiv.org/pdf/1606.03126v1.pdf) 14 | - [Semantic Parsing via Staged Query Graph Generation: Question Answering with Knowledge Base](https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/ACL15-STAGG.pdf) 15 | - [Value of Semantic Parse Labeling for KBQA](https://www.microsoft.com/en-us/research/wp-content/uploads/2016/08/acl2016-webqsp.pdf) 16 | - [Question Answering with Subgraph Embeddings](https://arxiv.org/pdf/1406.3676v3.pdf) 17 | - [Open Question Answering with Weakly Supervised Embedding Models](https://arxiv.org/pdf/1404.4326.pdf) 18 | - [Learning End-to-End Goal-Oriented dialog](https://arxiv.org/pdf/1605.07683v2.pdf) 19 | - [End-to-End Memory Networks with Knowledge Carryover for Multi-Turn Spoken Language Understanding](https://www.microsoft.com/en-us/research/wp-content/uploads/2016/06/IS16_ContextualSLU.pdf) 20 | - [Question Answering over Knowledge Base With Neural Attention Combining Global Knowledge Information](https://arxiv.org/pdf/1606.00979v1.pdf) 21 | - [Compositional Learning of Embeddings for Relation Paths in Knowledge Bases and Texts](https://www.microsoft.com/en-us/research/wp-content/uploads/2016/06/acl2016relationpaths-1.pdf) 22 | - [Neural Machine Translation by jointly learning to align and translate](https://arxiv.org/pdf/1409.0473v7.pdf) 23 | - [Recurrent Neural Network Grammar](https://arxiv.org/pdf/1602.07776v4.pdf) 24 | - [Neural Turing Machines](https://www.youtube.com/watch?v=_H0i0IhEO2g) 25 | - [Teaching machines to read and comprehend](https://arxiv.org/pdf/1506.03340.pdf) 26 | - [Applying Deep Learning to answer selection: A study and an open task](https://arxiv.org/pdf/1508.01585v2.pdf) 27 | - [Reasoning with Neural Tensor Networks](https://papers.nips.cc/paper/5028-reasoning-with-neural-tensor-networks-for-knowledge-base-completion.pdf) 28 | - [Scalable Feature Learning for networks: Node2Vec](https://cs.stanford.edu/people/jure/pubs/node2vec-kdd16.pdf) 29 | - [Learning Distributed Representations for Rooted Subgraphs from Large Graphs: Subgraph2Vec](https://arxiv.org/pdf/1606.08928.pdf) 30 | - [Hybrid computing using a neural network with dynamic external memory](http://www.nature.com/nature/journal/v538/n7626/full/nature20101.html) 31 | - [Traversing Knowledge Graphs in Vector Space](http://www.emnlp2015.org/proceedings/EMNLP/pdf/EMNLP038.pdf) 32 | - [Learning to Compose Neural Networks for Question Answering](https://arxiv.org/abs/1601.01705) 33 | - [Hierarchical Memory Networks](http://openreview.net/pdf?id=BJ0Ee8cxx) 34 | - [Gaussian Attention Model and its Application to Knowledge Base Embedding and Question Answering](https://arxiv.org/pdf/1611.02266.pdf) 35 | - [Gated Graph Sequence Neural Networks](https://arxiv.org/abs/1511.05493) 36 | - [Sequence to Sequence Learning With Neural Networks](https://papers.nips.cc/paper/5346-sequence-to-sequence-learning-with-neural-networks.pdf) 37 | - [Neural Conversation Model](https://arxiv.org/pdf/1506.05869v1.pdf) 38 | - [Query Reduction Networks For Question Answering](https://arxiv.org/pdf/1606.04582.pdf) 39 | - [Conditional Focused Neural Question Answering with Large-scale Knowledge Bases](https://arxiv.org/pdf/1606.01994.pdf) 40 | - [Efficiently Answering Technical Questions — A Knowledge Graph Approach](http://wangzhongyuan.com/en/papers/Technical_Questions_Answering.pdf) 41 | - [An End-to-End Model for Question Answering over Knowledge Base with Cross-Attention Combining Global Knowledge](http://www.nlpr.ia.ac.cn/cip/~liukang/liukangPageFile/ACL2017-Hao.pdf) 42 | - [Question Answering as Global Reasoning over Semantic Abstractions](http://www.cis.upenn.edu/~danielkh/files/2018_semanticilp/2018_aaai_semanticilp.pdf) 43 | 44 | #### Category 45 | ##### Question generation 46 | - [Question Generation via Overgenerating Transformations and Ranking (Technical report)](https://www.lti.cs.cmu.edu/sites/default/files/cmulti09013.pdf) 47 | - [Automation of question generation from sentences](http://www.sadidhasan.com/sadid-QG.pdf) 48 | - [Good question!statistical ranking for question generation](https://homes.cs.washington.edu/~nasmith/papers/heilman+smith.naacl10.pdf) 49 | - [Question generation from paragraphs at upenn: Qgstec system description](http://www.aclweb.org/anthology/I11-1104) 50 | - [Automatically generating questions from queries for community-based question answering](http://www.aclweb.org/anthology/I11-1104) 51 | - [How to Generate Cloze Questions from Definitions: A Syntactic Approach](https://www.cs.cmu.edu/~listen/pdfs/gates-2011-aaai-qg.pdf) 52 | - [Generating natural language questions to support learning on-line](http://www.aclweb.org/anthology/W13-2114) 53 | - [Deep questions without deep understanding](http://www.aclweb.org/anthology/P15-1086) 54 | - [Leveraging multiple views of text for automatic question generation](http://link.springer.com/chapter/10.1007/978-3-319-19773-9_26) 55 | - [Revup: Automatic gap-fill question generation from educational texts](http://www.aclweb.org/anthology/W15-0618) 56 | - [Towards topic-to-question generation](http://www.mitpressjournals.org/doi/abs/10.1162/COLI_a_00206) 57 | - [Ranking automatically generated questions using common human queries](http://www.aclweb.org/old_anthology/W/W16/W16-66.pdf#page=233) 58 | - [Generating quiz questions from knowledge graphs](http://delivery.acm.org/10.1145/2750000/2742722/p113-seyler.pdf) 59 | - [Generating Factoid Questions With Recurrent Neural Networks: The 30M Factoid Question-Answer Corpus](http://arxiv.org/pdf/1603.06807v1.pdf) 60 | - [Knowledge Questions from Knowledge Graphs](https://arxiv.org/abs/1610.09935) 61 | - [Machine Comprehension by Text-to-Text Neural Question Generation](http://aclweb.org/anthology/W17-2603) 62 | - [Question Generation from a Knowledge Base with Web Exploration](https://arxiv.org/pdf/1610.03807.pdf) 63 | - [On Generating Characteristic-rich Question Sets for QA Evaluation](http://www.aclweb.org/anthology/D/D16/D16-1054.pdf) 64 | - [Neural Question Generation from Text: A Preliminary Study](https://arxiv.org/pdf/1704.01792.pdf) 65 | - [Semi-supervised qa with generative domain-adaptive nets](https://pdfs.semanticscholar.org/e8a0/536dc080acd2ca83502dddd0d511ef3fbd8c.pdf) 66 | 67 | 68 | #### Datasets 69 | - [bAbI dataset](https://research.facebook.com/research/babi/) 70 | - [CNN QA Task (Teaching Machines to Read & Comprehend)](https://github.com/deepmind/rc-data/) 71 | - [WebQuestions](http://nlp.stanford.edu/software/sempre/) 72 | - [Simple Questions](https://research.facebook.com/research/babi) 73 | - [Movie QA](https://research.facebook.com/research/babi/) 74 | - [WebQuestionsSP](https://www.microsoft.com/en-us/download/details.aspx?id=52763) 75 | - [WikiQA](https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/YangYihMeek_EMNLP-15_WikiQA.pdf) 76 | - [Kaggle AllenAI Challenge](https://www.kaggle.com/c/the-allen-ai-science-challenge) 77 | - [MC Test, Machine Comprehension Test Microsoft 2013](http://research.microsoft.com/en-us/um/redmond/projects/mctest/) 78 | - [MSR Sentence Completion Challenge](https://www.microsoft.com/en-us/research/project/msr-sentence-completion-challenge/) 79 | - [Dialog State Tracking Challenge](http://camdial.org/~mh521/dstc/) 80 | - [QA dataset featured in Teaching Machines to Read and Comprehend](https://github.com/deepmind/rc-data/) 81 | - [WebNav](https://github.com/nyu-dl/WebNav/blob/master/README.md) 82 | - [Stanford Question Answering Dataset](https://rajpurkar.github.io/SQuAD-explorer/) 83 | - [FB15K Knowledge Base](https://www.microsoft.com/en-us/download/details.aspx?id=52312) 84 | - Yahoo! Answers Comprehensive Questions and Answers version 1.0 (multi part) 85 | - [Cornell Movie Dialogue Dataset](https://www.cs.cornell.edu/~cristian/Cornell_Movie-Dialogs_Corpus.html) 86 | - [WikiQA](http://aka.ms/WikiQA) 87 | - [Quora Duplicate Questions Dataset](https://data.quora.com/) 88 | - [Query Reformulator Dataset Jeopardy etc](https://github.com/nyu-dl/QueryReformulator) 89 | - [Quiz Bowl Questions](https://www.cs.colorado.edu/~jbg/projects/IIS-1320538.html#Datasets) 90 | - [WebQA-Chinese](http://idl.baidu.com/WebQA.html) 91 | - [Chat corpus](https://github.com/Marsan-Ma/chat_corpus) 92 | - [MultiRC](http://cogcomp.org/multirc/) 93 | 94 | #### KBs 95 | - [NetBase](https://github.com/pannous/netbase) 96 | - [Freebase](https://developers.google.com/freebase/) 97 | 98 | #### Presentations 99 | - [Relation Learning for Large Scale Knowledge Graph](http://nlp.csai.tsinghua.edu.cn/~lzy/talks/adl2015.pdf) 100 | - [Attention and Memory](http://videolectures.net/site/normal_dl/tag=1051694/deeplearning2016_chopra_attention_memory_01.pdf) 101 | 102 | #### LM Datasets 103 | - PennTree Bank 104 | - Text8 105 | 106 | #### Code & Relevant Projects 107 | - [MemNN Impl Matlab](https://github.com/facebook/MemNN) 108 | - [Key Value MemNN](https://github.com/siyuanzhao/key-value-memory-networks) 109 | - [Quepy](https://github.com/machinalis/quepy) 110 | - [NLQuery](https://github.com/ayoungprogrammer/nlquery) 111 | - [ParlAI](https://github.com/facebookresearch/ParlAI) 112 | - [flask-chatterbot](https://github.com/chamkank/flask-chatterbot) 113 | - [Learning to Rank short text pairs with CNN SIGIR 2015](https://github.com/shashankg7/Keras-CNN-QA) 114 | - [TextKBQA](https://github.com/rajarshd/TextKBQA) 115 | - [BiAttnFlow](https://github.com/allenai/bi-att-flow) 116 | -------------------------------------------------------------------------------- /ReadingNotes.md: -------------------------------------------------------------------------------- 1 | # QA Readings 2 | ## [Memory Networks][memNet] 3 | 4 | * Training 5 | * Supervised by giving outputs for max function in 'O' and 'R' 6 | * Train embedding matrices using SGD and margin loss ranking 7 | 8 | ## [End-to-End Memory Networks][EToEMemNet] 9 | 10 | ### [**Source Code][EToEMemNetSource] 11 | 12 | ## [Towards AI-Complete QA: A set of Prerequisite Toy Tasks][toyTasks] 13 | 14 | * Dataset is self contained. 15 | * Tasks come with both training and evaluation data 16 | * These tasks can be generated by source code found [here](https://research.facebook.com/research/babi/) 17 | 18 | ## [Large-scale Simple QA with Memory Networks][largeScale] 19 | 20 | * Training on multiple datasets had no negative results. Only improved performance 21 | * Datasets: 22 | * [Freebase](http://www.freebase.com) 23 | * [WebQuestions](http://nlp.stanford.edu/software/sempre/) 24 | * [Reverb](reverb.cs.washing.edu) 25 | * This site includes source code 26 | * [SimpleQuestions](https://research.facebook.com/research/babi) 27 | * New dataset contributed by this paper 28 | 29 | ## [Ask Me Anything: Dynamic Memory Networking for NLP][AMA] 30 | 31 | * Datasets 32 | * Used the [toy tasks](https://research.facebook.com/research/babi/) mentioned above 33 | * Concluded that Dynamic Memory Networks did well for general NLP structure 34 | 35 | ## [Key-Value Memory Networks for Directly Reading Documents][Key-Value] 36 | 37 | * New approach of Key-Value MemNetworks made reading documents more viable 38 | * Datasets: 39 | * [MovieQA](https://research.facebook.com/research/babi/) ~ 100k questions in movie domain 40 | * Knowledge Base formed from data available on [OmbdApi](http://omdbapi.com/) 41 | 42 | ## [Semantic Parsing via Staged Query Graph Generation][semanticGraph] 43 | 44 | * Datasets 45 | * [WebQuestions](http://nlp.stanford.edu/software/sempre/) mentioned above 46 | * Evaluated results with the [SEMPRE](http://www-nlp.stanford.edu/software/sempre/) 47 | * Semantic parsing is reduced to graph query generation, formulated as a staged search problem 48 | * Leverages KB at an early stage to prune search space 49 | 50 | ## [The Value of Semantic Parse Labeling for KBQA][semanticValue] 51 | 52 | * Datasets 53 | * [WebQuestionsSP](https://www.microsoft.com/en-us/download/details.aspx?id=52763) 54 | * Learning from labeled semantic parses significantly improves overall performance 55 | * Can obtain semantic parses with high accuracy and at a cost comparable to obtaining just the answers 56 | 57 | ## [QA with Subgraph Enbeddings][subgraph] 58 | 59 | * Datasets 60 | * Trained with [FreeBase](http://www.freebase.com) 61 | * Tested on [WebQuestions](http://nlp.stanford.edu/software/sempre/) 62 | * Very weak performance, peaked at 41.8 F1 score, but the best (so far) with no human interaction. 63 | 64 | ## [Open QA with Weakly Supervised Embedding Models][openWeak] 65 | 66 | * Databases 67 | * [DBPedia](http://wiki.dbpedia.org/) 68 | * [WebQuestions](http://nlp.stanford.edu/software/sempre/) 69 | * Learn to map questions to vectorial feature representations 70 | * Allows use of any KB without a required lexicon or grammar 71 | * All variations greatly improved upon [Paralex](http://www.aclweb.org/anthology/P13-1158) 72 | 73 | [//]: # (These are reference links used in the body of this note and get stripped out when the markdown processor does its job. There is no need to format nicely because it shouldn't be seen. Thanks SO - http://stackoverflow.com/questions/4823468/store-comments-in-markdown-syntax) 74 | 75 | [memNet]: 76 | [EToEMemNet]: 77 | [EToEMemNetSource]: 78 | [toyTasks]: 79 | [largeScale]: 80 | [AMA]: 81 | [Key-Value]: 82 | [semanticGraph]: 83 | [semanticValue]: 84 | [subgraph]: 85 | [openWeak]: 86 | 87 | --------------------------------------------------------------------------------