1 / 0

Discourse Parsing in the Penn Discourse Treebank: Using Discourse Structures to Model Coherence and Improve User Tasks

Ph.D. Thesis Proposal. Discourse Parsing in the Penn Discourse Treebank: Using Discourse Structures to Model Coherence and Improve User Tasks. Ziheng Lin. Advisors: Prof Min-Yen Kan and Prof Hwee Tou Ng. Introduction. A text is usually understood by its discourse structure

marlie
Download Presentation

Discourse Parsing in the Penn Discourse Treebank: Using Discourse Structures to Model Coherence and Improve User Tasks

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Ph.D. Thesis Proposal

    Discourse Parsing in the Penn Discourse Treebank: Using Discourse Structures to Model Coherence and Improve User Tasks

    Ziheng Lin Advisors: Prof Min-Yen Kan and Prof HweeTou Ng
  2. Introduction A text is usually understood by its discourse structure Discourse parsing: a process of Identifying discourse relations, and Constructingthe internal discourse structure A number of discourse frameworks has been proposed: Mann & Thompson (1988) Lascarides & Asher (1993) Webber (2004) …
  3. Introduction The Penn Discourse Treebank (PDTB): Is a large-scale discourse-level annotation Follows Webber’s framework Understanding a text’s discourse structure is useful: Discourse structure and textual coherence have a strong connection Discourse parsing is useful in modeling coherence Discourse parsing also helps downstream NLP applications Contrast, Restatement  summarization Cause  QA
  4. Introduction Research goals: Design an end-to-end PDTB-styled discourse parser Propose a coherence model based on discourse structures Show discourse parsing improves downstream NLP application
  5. Outline Introduction Literature review Discourse parsing Coherence modeling Recognizing implicit discourse relations A PDTB-styled end-to-end discourse parser Modeling coherence using discourse relations Proposed work and timeline Conclusion
  6. Discourse parsing Recognize the discourse relations between two text spans, and Organize these relations into a discourse structure Two main classes of relations in PDTB: Explicit relations: explicit discourse connective such as however and because Implicit relations: no discourse connective, harder to recognize  parsing implicit relations is a hard task
  7. Discourse parsing Marcu & Echihabi (2002): Word pairs extracted from two text spans Collect implicit relations by removing connectives Wellner et al. (2006): Connectives, distance between text spans, and event-based features Discourse Graphbank: explicit and implicit Soricut & Marcu (2003): Probabilistic models on sentence-level segmentation and parsing RST Discourse Treebank (RST-DT) duVerle & Prendinger (2009): SVM to identify discourse structure and label relation types RST-DT Wellner & Pustejovsky (2007), Elwell & Baldridge (2008), Wellner (2009)
  8. Coherence modeling Barzilay & Lapata (2008): Local coherence Distribution of discourse entities exhibits certain regularities on a sentence-to-sentence transition Model coherence using an entity grid Barzilay & Lee (2004): Global coherence Newswire reports follow certain patterns of topic shift Used a domain-specific HMM model to capture topic shift in a text
  9. Outline Introduction Literature review Recognizing implicit discourse relations Methodology Experiments A PDTB-styled end-to-end discourse parser Modeling coherence using discourse relations Proposed work and timeline Conclusion
  10. Methodology Supervised learning on a maximum entropy classifier Four feature classes Contextual features Constituent parse features Dependency parse features Lexical features
  11. Methodology: Contextual features Dependencies between two adjacent discourse relations r1 and r2 independent fully embedded argument shared argument properly contained argument pure crossing partially overlapping argument Fully embedded argument and shared argument are the most common ones in the PDTB
  12. Methodology:Contextual features For an implicit relation currthat we want to classify, look at the surrounding two relations prev and next six binary features:
  13. Methodology:Constituent parse features Collect all production rules Three binary features to check whether a rule appears in Arg1, Arg2, and both S  NP VP NP  PRP PRP  “We” ……
  14. Methodology:Dependency parse features Encode additional information at the word level Collect all words with the dependency types from their dependents: Three binary features to check whether a rule appears in Arg1, Arg2, and both “had”  nsubj dobj “problems”  det nn advmod “at”  dep
  15. Methodology:Lexical features Marcu & Echihabi (2002) show word pairs are a good signal to classify discourse relations Arg1: John is good in math and sciences. Arg2: Paul fails almost every class he takes. (good, fails) is a good indicator for a contrast relation Stem and collect all word pairs from Arg1 and Arg2 as features
  16. Outline Introduction Literature review Recognizing implicit discourse relations Methodology Experiments A PDTB-styled end-to-end discourse parser Modeling coherence using discourse relations Proposed work and timeline Conclusion
  17. Experiments w/ feature selection Employed MI to select the top 100 rules, and top 500 word pairs (as word pairs are more sparse) Production rules, dependency rules, and word pairs all gave significant improvement with p < 0.01 Applying all feature classes yields the highest accuracy of 40.2% Results show predictiveness of feature classes: production rules > word pairs > dependency rules > context features
  18. Experiments Question: can any of these feature classes be omitted to achieve the same level of performance? Add in feature classes in the order of their predictiveness production rules > word pairs > dependency rules > context features The results confirm that each additional feature class contributes a marginal performance improvement, and all feature classes are needed for the optimal performance Production Dependency Word pairs Context Acc. Rules Rules 100 100 500 39.0% 100 100 500 Yes 40.2% 100 500 38.9% 100 38.4%
  19. Conclusion Implemented an implicit discourse relation classifier Features include: Modeling of the context of the relations Features extracted from constituent and dependency trees Word pairs Achieved an accuracy of 40.2%, a 14.1% improvement over the baseline With a component that handles implicit relations, continue to design a full parser
  20. Outline Introduction Literature review Recognizing implicit discourse relations A PDTB-styled end-to-end discourse parser System overview Components Experiments Modeling coherence using discourse relations Proposed work and timeline Conclusion
  21. System overview The parsing algo mimics the PDTB annotation procedure Input – a free text T Output – discourse structure of T in the PDTB style Three steps: Step 1: label Explicit relation Step 2: label Non-Explicit relation (Implicit, AltLex, EntRel and NoRel) Step 3: label attribution spans
  22. System overview
  23. Outline Introduction Literature review Recognizing implicit discourse relations A PDTB-styled end-to-end discourse parser System overview Components Experiments Modeling coherence using discourse relations Proposed work and timeline Conclusion
  24. Components:Connective classifier Use syntactic features from Pitler & Nenkova (2009) A connective’s context and POS give indication of its discourse usage E.g., after is a discourse connective when it is followed by a present particle, such as “after rising 3.9%” New contextual features for connective C: C POS prev + C, prev POS, prev POS + C POS C + next, next POS, C POS + next POS The path from C to the root
  25. Components:Argument labeler Label Arg1 and Arg2 spans in two steps: Step 1: identify the locations of Arg1 and Arg2 Step 2: label their spans Step 1 - argument position classifier: Arg2 is always associated with the connective Use contextual and lexical info to locate Arg1 Step 2 – argument extractor: Case 1 – Arg1 and Arg2 in the same sentence Case 2 – Arg1 in some previous sentence: assume the immediately previous
  26. Components: Explicit classifier Human agreement: 94% on Level-1 84% on Level-2 We train and test on Level-2 types Features: Connective C C POS C + prev
  27. Components: Non-Explicit classifier Non-Explicit: Implicit, AltLex, EntRel, NoRel Modify the implicit classifier to include the AltLex, EntRel and NoRel AleLex is signaled by non-connective expressions such as “That compared with”, which usually appear at the beginning of Arg2 Add another three features to check the beginning three words of Arg2
  28. Components: Attribution span labeler Label the attribution spans for Explicit, Implicit, and AltLex Consists of two steps: Step 1: split the text into clauses Step 2: decide which clauses are attribution spans Features from curr, prev and next clauses: Unigrams of curr Lowercased and lemmatized verbs in curr First term of curr, Last term of curr, Last term of prev, First term of next Last term of prev + first term of curr, Last term of curr + first term of next Position of curr in the sentence Production rules extracted from curr prev curr next
  29. Outline Introduction Literature review Recognizing implicit discourse relations A PDTB-styled end-to-end discourse parser System overview Components Experiments Modeling coherence using discourse relations Proposed work and timeline Conclusion
  30. Experiments Each component in the pipeline can be tested with two dimensions: Whether there is error propagation from previous component (EP vs no EP), and Whether gold standard parse trees and sentence boundaries or automatic parsing and sentence splitting are used (GS vs Auto) Three settings: GS + no EP: per component evaluation GS + EP Auto + EP: fully automated end-to-end evaluation
  31. Experiments Connective classifier Argument extractor
  32. Experiments Explicit classifier Non-explicit classifier
  33. Experiments Attribution span labeler Evaluate the whole pipeline: GS + EP gives F1 of 46.8% under partial match and 33% under exact match Auto + EP gives F1 of 38.18% under partial match and 20.64% under exact match
  34. Conclusion Designed and implemented an end-to-end PDTB-styled parser Incorporated the implicit classifier into the pipeline Evaluated the system both component-wise as well as with error propagation Reported overall system F1 for partial match of 46.8% with gold standard parses and 38.18% with full automation
  35. Outline Introduction Literature review Recognizing implicit discourse relations A PDTB-styled end-to-end discourse parser Modeling coherence using discourse relations A relation transition model A refined approach: discourse role matrix Conclusion Proposed work and timeline Conclusion
  36. A relation transition model Recall: Barzilay & Lapata (2008)'s coherence representation models sentence-to-sentence transitions of entities Well-written texts follow certain patterns of argumentative moves Reflected by relation transition patterns A text T can be represented as a relation transition:
  37. A relation transition model Method and preliminary results: Extract the relation bigrams from the relation transition sequence [Cause Cause], [Cause Contrast], [Contrast Restatement], [Restatement Expansion] A training/test instance is a pair of relation sequences: Sgs = gold standard sequence Sp = permuted sequence Task: rank the pair (Sgs, Sp) Ideally, Sgs should be ranked higher, ie, more coherent Baseline: 50%
  38. A relation transition model The rel transition sequence is sparse Expect longer articles to give more predictable sequence Perform experiments with diff sentence thresholds
  39. Outline Introduction Literature review Recognizing implicit discourse relations A PDTB-styled end-to-end discourse parser Modeling coherence using discourse relations A relation transition model A refined approach: discourse role matrix Conclusion Proposed work and timeline Conclusion
  40. A refined approach: discourse role matrix Instead of looking at the discourse roles of sentences, we look at the discourse roles of terms Use sub-sequences of discourse roles as features Comp.Arg2  Exp.Arg2, Comp.Arg1  nil, …
  41. A refined approach: discourse role matrix Experiments: Compared with Barzilay & Lapata (2008) ’s entity grid model
  42. Outline Introduction Literature review Recognizing implicit discourse relations A PDTB-styled end-to-end discourse parser Modeling coherence using discourse relations Proposed work and timeline Literature review on several NLP applications Proposed work Timeline Conclusion
  43. Literature review on several NLP applications Text summarization: Discourse plays an important role in text summarization Marcu (1997) showed that RST tree is a good indicator of salience in text PDTB relations are helpful in summarization: Generic summarization: utilize Instantiation and Restatement relations to recognize redundancy Update summarization: use Contrast relations to locate updates
  44. Literature review on several NLP applications Argumentative zoning (AZ): Proposed by Teufel (1999) to automatically construct the rhetorical moves of argumentation of academic writings Label sentences with 7 tags: aim, textual, own, background, contrast, basis, and other Has been shown that AZ can help in: Summarization (Teufel & Moens, 2002) Citation indexing (Teufel et al., 2006)
  45. Literature review on several NLP applications Why-QA: Aims to answer generic question “Why X?” Verberne et al. (2007) showed that discourse structure in RST framework is helpful in a why-QA system Prasad and Joshi (2008) generate why-questions with the use of causal relations in the PDTB We believe that the PDTB hierarchical relation typing will help in designing a why-QA system
  46. Proposed work Work done: A system to automatically recognize implicit relations Sec 3, EMNLP 2009 An end-to-end discourse parser Sec 4, a journal in preparation Coherence model based on discourse structures Sec 5, ACL 2011 Next step, I propose to work on one of the NLP applications Aim: show that discourse parsing can improve the performance of this NLP app
  47. Timeline
  48. Outline Introduction Literature review Recognizing implicit discourse relations A PDTB-styled end-to-end discourse parser Modeling coherence using discourse relations Proposed work and timeline Conclusion
  49. Conclusion Designed and implemented an implicit discourse classifier in the PDTB Designed and implemented an end-to-end discourse parser in the PDTB representation Proposed a coherence model based on discourse relations Proposed work: apply discourse parsing in one downstream NLP application Summarization, argumentative zoning, or why-QA Parser Demo
  50. Thank you!
  51. Back up slides
  52. The Penn Discourse Treebank A discourse level annotation over the WSJ corpus Adopts a binary predicate-argument view on discourse relations Explicit relations: signaled by discourse connectives Arg2:When he sent letters offering 1,250 retired major leaguers the chance of another season, Arg1: 730 responded. Implicit relations: Arg1: “I believe in the law of averages,” declared San Francisco batting coach Dusty Baker after game two. Arg2: [accordingly] “I’d rather see a so-so hitter who’s hot come up for the other side than a good hitter who’s cold.”
  53. The Penn Discourse Treebank AltLex relations: Arg1: For the nine months ended July 29, SFE Technologies reported a net loss of $889,000 on sales of $23.4 million. Arg2:AltLex [That compared with] an operating loss of $1.9 million on sales of $27.4 million in the year-earlier period. EntRel: Arg1: Pierre Vinken, 61 years old, will join the board as a nonexecutive director Nov. 29. Arg2: Mr. Vinken is chairman of Elsevier N.V., the Dutch publishing group.
  54. The Penn Discourse Treebank
  55. Experiments Classifier: OpenNLPMaxEnt Training data: Sections 2 – 21 of the PDTB Test data: Section 23 of the PDTB Feature selection: Use Mutual Information(MI) to select features for production rules, dependency rules, and word pairs separately Majority baseline: 26.1%, where all instances are classified into Cause
  56. Components:Argument labeler
  57. A relation transition model can be represented by: or:
  58. Experiments The classifier labels no instances of Synchrony, Pragmatic Cause, Concession, and Alternative The percentages of these four types are too small: totally only 4.76% in the training data As Cause is the most predominant type, it has high recall but low precision
  59. Methodology:Constituent parse features Syntactic structure within one argument may constrain the relation type and the syntactic structure of the other argument (a) Arg1: But the RTC also requires “working” capital to maintain the bad assets of thrifts that are sold Arg2: [subsequently] That debt would bepaid off as the assets are sold (b) Arg1: It would have been too late to think about on Friday. Arg2: [so] We had to think about it ahead of time.
  60. Components:Connective classifier PDTB defines 100 discourse connectives Features from Pitler and Nenkova (2009): Connective: because Self category: IN Parent category: SBAR Left sibling category: none Right sibling category: S Right sibling contains a VP: yes Right sibling contains a trace: no trace
  61. Experiments Connective classifier: Adding the lexico-syntactic and path features significantly (p < 0.001) improves accuracy and F1 for both GS and Auto The connective with the highest number of incorrect labels is and and is always regarded as an ambiguous connective
  62. Experiments Argument position classifier: Performance drops when EP and Auto are added in The degradation is mostly due to the SS class False positives propagated from connective classifier For GS + EP: 30/36 classified as SS For Auto + EP: 46/52 classified as SS  the difference between SS and PS is largely due to error propagation
  63. Experiments Argument extractor - argument node identifier: F1 for Arg1, Arg2, and Rel (Arg1+Arg2) Arg1/Arg2 nodes for subordinating connectives are the easiest ones to locate 97.93% F1 for Arg2, 86.98% F1 for Rel Performance for discourse adverbials are the lowest Their Arg1 and Arg2 nodes are not strongly bound
  64. Experiments Argument extractor: Report both partial and exact match GS + no EP gives a satisfactory Rel F1 of 86.24% for partial match The performance for exact match is much lower than human agreement (90.2%) Most misses are due to small portions of text being deleted from / added to the spans by the annotators
  65. Experiments Explicit classifier: Human agreement = 84% A baseline that uses only connective as features yields an F1 of 86% under GS + no EP Adding new features improves to 86.77%
  66. Experiments Non-explicit classifier: A majority baseline (all classified as EntRel) gives F1 in the low 20s GS + no EP shows a F1 of 39.63% Performance for GS + EP and Auto + EP are much lower Still outperforms baseline by ~6%
  67. Experiments Attribution span labeler: GS + no EP achieves F1 of 79.68% and 65.95% for partial and exact match With EP: degradation is mostly due to the drop in precision With Auto: degradation is mostly due to the drop in recall
  68. Experiments Evaluate the whole pipeline: Look at the Explicit and Non-Explicit relations that are correctly identified Define a relation as correct if its relation type is classified correctly, and both Arg1 and Arg2 are labeled correctly (partial or exact) GS + EP gives F1 of 46.8% under partial match and 33% under exact match Auto + EP gives F1 of 38.18% under partial match and 20.64% under exact match A large portion of misses come from the Non-Explicit relations
  69. A lexical model Lapata (2003) proposed a sentence ordering model Assume the coherence of adjacent sentences is based on lexical word pairs: The coherence of the text is thus: RST enforces two possible canonical orders of text spans: Satellite before nucleus (e.g., conditional) Nucleus before satellite (e.g., restatement) A word pair-based model can be used to check whether these orderings are enforced
  70. A lexical model Method and preliminary results: Extract (wi-1,j, C, wi,k) as features: Use mutual information to select top n features, n = 5000 Accuracy = 70%, baseline = 50%
  71. Experiments w/o feature selection Production rules and word pairs yield significantly better performance Contextual features perform slightly better than the baseline Dependency rules perform slightly lower than baseline, and applying all feature classes does not yield the highest accuracy  noise
  72. Components: Argument labeler: Argument position classifier Relative positions of Arg1: SS: in the same sentence as the connective (60.9%) PS: in some previous sentence of the connective (39.1%) FS: in some sentence following the sentence of the connective (0%, only 8 instances, thus ignored) Classify the relative position of Arg1 as SS or PS Features: Connective C, C POS Position of C in the sentence (start, middle, end) prev1, prev1 POS, prev1 + C, prev1 POS + C POS prev2, prev2 POS, prev2 + C, prev2 POS + C POS
  73. Components: Argument labeler: Argument extractor When Arg1 is classified as in the same sentence (SS) as Arg2, it can be one of: Arg1 before Arg2 Arg2 before Arg1 Arg1 embedded within Arg2 Arg2 embedded within Arg1 Arg1 and Arg2 nodes in the parse tree can be syntactically related in one of three ways:
  74. Components: Argument labeler: Argument extractor Design an argument node identifier to identify the Arg1 and Arg2 subtree nodes within the sentence parse tree Features: Connective C C’s syntactic category (subordinate, coordinate, adverbial) Numbers of left and right siblings of C Path P of C to the node under consideration Path P and whether the size of C’s left sibling is greater than one The relative position of the node to C
  75. Components: Argument labeler: Argument extractor When Arg1 is classified as in some previous sentence (PS), we use the majority classifier Label the immediately previous sentence as Arg1 (76.9%)
More Related