1 / 50

Containment, Exclusion, and Implicativity: A Model of Natural Logic for Textual Inference

Containment, Exclusion, and Implicativity: A Model of Natural Logic for Textual Inference. Bill MacCartney and Christopher D. Manning NLP Group Stanford University 14 February 2008.

winthrop
Download Presentation

Containment, Exclusion, and Implicativity: A Model of Natural Logic for Textual Inference

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. Containment, Exclusion, and Implicativity:A Model of Natural Logic for Textual Inference Bill MacCartney and Christopher D. Manning NLP Group Stanford University 14 February 2008

  2. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion The textual inference task • Does premise P justify an inference to hypothesis H? • An informal, intuitive notion of inference: not strict logic • Focus on local inference steps, not long chains of deduction • Robust, accurate textual inference could enable: • Question answering [Harabagiu & Hickl 06] • Semantic search • Customer email response • Relation extraction (database building) • Document summarization • A limited selection of data sets: RTE, FraCaS, KBE

  3. OK No western state completely forbids casino gambling. No state completely forbids gambling. Few or no states completely forbid casino gambling. No No state completely forbids casino gambling for kids. No state restricts gambling. No state or city completely forbids casino gambling. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Some simple inferences No state completely forbids casino gambling. What kind of textual inference system could predict this?

  4. Bos & Markert 2006 FOL &theoremproving semantic graph matching Hickl et al. 2006 MacCartney et al. 2006 Burchardt & Frank 2006 patternedrelationextraction Romano et al. 2006 lexical/semanticoverlap Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Jijkoun & de Rijke 2005 Textual inference:a spectrum of approaches deep,but brittle naturallogic robust,but shallow

  5. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion What is natural logic? • (natural logic  natural deduction) • Lakoff (1970) defines natural logic as a goal (not a system) • to characterize valid patterns of reasoning via surface forms (syntactic forms as close as possible to natural language) • without translation to formal notation:       • A long history • traditional logic: Aristotle’s syllogisms, scholastics, Leibniz, … • van Benthem & Sánchez Valencia (1986-91): monotonicity calculus • Precise, yet sidesteps difficulties of translating to FOL: idioms, intensionality and propositional attitudes, modalities, indexicals, reciprocals,scope ambiguities, quantifiers such as most, reciprocals, anaphoric adjectives, temporal and causal relations, aspect, unselective quantifiers, adverbs of quantification, donkey sentences, generic determiners, …

  6. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion + – + – – – + + + Monotonicity calculus (Sánchez Valencia 1991) • Entailment as semantic containment: rat< rodent, eat< consume, this morning< today, most< some • Monotonicity classes for semantic functions • Upward monotone: some rats dream< some rodents dream • Downward monotone: no rats dream > no rodents dream • Non-monotone: most rats dream # most rodents dream • Handles even nested inversions of monotonicity Every state forbids shooting game without a hunting license • But lacks any representation of exclusion Garfield is a cat< Garfield is not a dog

  7. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Implicatives & factives • Work at PARC, esp. Nairn et al. 2006 • Explains inversions & nestings of implicatives & factives • Ed did not forget to force Dave to leave  Dave left • Defines 9 implication signatures • “Implication projection algorithm” • Bears some resemblance to monotonicity calculus • But, fails to connect to containment or monotonicity • John refused to danceJohn didn’t tango

  8. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Outline • Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion

  9. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion A theory of natural logic Three elements: • an inventory of entailment relations • semantic containment relations of Sánchez Valencia • plus semantic exclusion relations • a concept of projectivity • explains entailments compositionally • generalizes Sánchez Valencia’s monotonicity classes • generalizes Nairn et al.’s implication signatures • a weak proof procedure • composes entailment relations across chains of edits

  10. Yesentailment Nonon-entailment 2-wayRTE1,2,3 Yesentailment Unknownnon-entailment Nocontradiction 3-wayFraCaS, PARC,RTE4? Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion P = Qequivalence P < Qforwardentailment P > Qreverseentailment P # Qnon-entailment containmentSánchez-Valencia Entailment relations in past work

  11. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion 16 elementary set relations Q Q P P P and Q can representsets of entities (i.e., predicates)or of possible worlds (propositions)cf. Tarski’s relation algebra

  12. P ^ Q P _ Q P > Q P = Q Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion P < Q P | Q P # Q 16 elementary set relations Q Q P P P and Q can representsets of entities (i.e., predicates)or of possible worlds (propositions)cf. Tarski’s relation algebra

  13. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion 7 basic entailment relations

  14. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Relations for all semantic types The entailment relations are defined for expressions of all semantic types (not just sentences)

  15. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Monotonicity of semantic functions Sánchez Valencia’s monotonicity calculus assigns semantic functions to one of three monotonicity classes: Upward-monotone (M) The default: “bigger” inputs yield “bigger” outputs Example: broken. Since chairfurniture, broken chairbroken furniture Heuristic: in a M context, broadening edits preserve truth Downward-monotone (M) Negatives, restrictives, etc.: “bigger” inputs yield “smaller” outputs Example: doesn’t. While hoverfly, doesn’t flydoesn’t hover Heuristic: in a M context, narrowing edits preserve truth Non-monotone (#M) Superlatives, some quantifiers (most, exactly n): neither M nor M Example: most. While penguinbird, most penguins # most birds Heuristic: in a #M context, no edits preserve truth

  16. explicit negation:no, n’t restrictive quantifiers:no, few, at most n didn’t dance< didn’t tango at most two people< at most two men negative & restrictive nouns:ban, absence [of], refusal negative & restrictive verbs:lack, fail, prohibit, deny drug ban< heroin ban prohibit weapons< prohibit guns Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion prepositions & adverbs:without, except, only the antecedent of a conditional If stocks rise, we’ll get real paid<If stocks soar, we’ll get real paid without clothes< without pants Downward monotonicity Downward-monotone constructions are widespread!

  17. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Generalizing to projectivity • How do the entailments of a compound expression depend on the entailments of its parts? • How does the entailment relation between (fx) and (fy) depend on the entailment relation between x and y(and the properties of f)? • Monotonicity gives partial answer (for =, <, >, #) • But what about the other relations (^, |, _)? • We’ll categorize semantic functions based on how they project the basic entailment relations

  18. downwardmonotonicity Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion swaps these too Example: projectivity of not

  19. downwardmonotonicity switch blocks, not swaps Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Example: projectivity of refuse

  20. @ @ > > > @ @ < < @ @ Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion @ @ nobody nobody can can without without a shirt clothes enter enter Projecting entailment relations upward Nobody can enter without a shirt < Nobody can enter without clothes • Assume idealized semantic composition trees • Propagate lexical entailment relations upward, according to projectivity class of each node on path to root

  21. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion A weak proof procedure • Find sequence of edits connecting P and H • Insertions, deletions, substitutions, … • Determine lexical entailment relation for each edit • Substitutions: depends on meaning of substituends • Deletions: < by default: red socks < socks • But some deletions are special: not hungry ^ hungry • Insertions are symmetric to deletions: > by default • Project up to find entailment relation across each edit • Compose entailment relations across sequence of edits

  22. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Composing entailment relations • Relation composition: if a R b and b S c, then a ? C • cf. Tarski’s relation algebra • Many compositions are intuitive = º =  = < º <  < < º =  < ^ º ^  = • Some less obvious, but still accessible | º ^  < fish | human, human ^ nonhuman,fish < nonhuman • But some yield unions of basic entailment relations! | º |  {=, <, >, |, #} (i.e. the non-exhaustive relations) • Larger unions convey less information (can approx. with #) • This limits power of proof procedure described 

  23. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Implicatives & factives • Nairn et al. 2006 define nine implication signatures • These encode implications (+, –, o) in + and – contexts • Refuse has signature –/o:refuse to danceimplies didn’t dancedidn’t refuse to dance implies neither danced nor didn’t dance • Signatures generate different relations when deleted • Deleting –/o generates |Jim refused to dance | Jim dancedJim didn’t refuse to dance _ Jim didn’t dance • Deleting o/– generates <Jim attempted to dance<Jim dancedJim didn’t attempt to dance>Jim didn’t dance • (Factives are only partly explained by this account)

  24. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Outline • Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion

  25. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion The NatLog system textual inference problem linguistic analysis 1 alignment 2 lexical entailment classification 3 entailment projection 4 entailment composition 5 prediction

  26. No S completely state VP forbid NP ADVP NP Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion DT NNS RB VBD NN NN casino gambling + – – – + + Step 1: Linguistic analysis • Tokenize & parse input sentences (future: & NER & coref & …) • Identify items w/ special projectivity & determine scope • Problem: PTB-style parse tree  semantic structure! no pattern: DT < /^[Nn]o$/ arg1: M on dominating NP __ >+(NP) (NP=proj !> NP) arg2: M on dominating S __ > (S=proj !> S) No state completely forbids casino gambling • Solution: specify scope in PTB trees using Tregex

  27. MAT INS MAT SUB DEL MAT Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Step 2: Alignment • Phrase-based alignments: symmetric, many-to-many • Can view as sequence of atomic edits: DEL, INS, SUB, MAT • Ordering of edits defines path through intermediate forms • Need not correspond to sentence order • Ordering of some edits can influence effective projectivity for others • We use heuristic reordering of edits to simplify this • Decomposes problem into atomic entailment problems • We haven’t (yet) invested much effort here • Experimental results use alignments from other sources

  28. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Running example OK, the example is contrived, but it compactly exhibits containment, exclusion, and implicativity

  29. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Step 3: Lexical entailment classification • Predict basic entailment relation for each edit, based solely on lexical features, independent of context • Feature representation: • WordNet features: synonymy, hyponymy, antonymy • Other relatedness features: Jiang-Conrath (WN-based), NomBank • String and lemma similarity, based on Levenshtein edit distance • Lexical category features: prep, poss, art, aux, pron, pn, etc. • Quantifier category features • Implication signatures (for DEL edits only) • Decision tree classifier • Trained on 2,449 hand-annotated lexical entailment problems • >99% accuracy on training data — captures relevant distinctions

  30. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Running example

  31. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Step 4: Entailment projection

  32. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion interesting final answer Step 5: Entailment composition

  33. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Outline • Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion

  34. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion The FraCaS test suite • FraCaS: mid-90s project in computational semantics • 346 “textbook” examples of textual inference problems • examples on next slide • 9 sections: quantifiers, plurals, anaphora, ellipsis, … • 3 possible answers: yes, no, unknown (not balanced!) • 55% single-premise, 45% multi-premise (excluded)

  35. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion FraCaS examples

  36. 27% error reduction in largest category, all but one correct high accuracy in sections most amenable to natural logic Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion high precision even outsideareas of expertise Results on FraCaS

  37. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion FraCaS confusion matrix guess gold

  38. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Outline • Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion

  39. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion The RTE3 test suite • RTE: more “natural” textual inference problems • Much longer premises: average 35 words (vs. 11) • Binary classification: yes and no • RTE problems not ideal for NatLog • Many kinds of inference not addressed by NatLog: paraphrase, temporal reasoning, relation extraction, … • Big edit distance  propagation of errors from atomic model

  40. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion RTE3 examples

  41. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Results on RTE3 data • Accuracy is unimpressive, but precision is relatively high • Maybe we can achieve high precision on a subset? • Strategy: hybridize with broad-coverage RTE system • As in Bos & Markert 2006 (each data set contains 800 problems)

  42. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion A simple bag-of-words model H P similarity scores on [0, 1]for each pair of words (I used a really simple-mindedsimilarity function based onLevenshtein string-edit distance) max sim for each hyp word how rare each word is = (max sim)^IDF = h P(h|P)

  43. +10probs Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Results on RTE3 data (each data set contains 800 problems)

  44. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Combining BoW & NatLog • MaxEnt classifier • BoW features: P(H|P), P(P|H) • NatLog features:7 boolean features encoding predicted entailment relation

  45. +11probs +3probs Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Results on RTE3 data (each data set contains 800 problems)

  46. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Problem: NatLog is too precise? • Error analysis reveals a characteristic pattern of mistakes: • Correct answer is yes • Number of edits is large (>5) (this is typical for RTE) • NatLog predicts < or = for all but one or two edits • But NatLog predicts some other relation for remaining edits! • Most commonly, it predicts > for an insertion (e.g., RTE3_dev.71) • Result of relation composition is thus #, i.e. no • Idea: make it more forgiving, by adding features • Number of edits • Proportion of edits for which predicted relation is not < or =

  47. +13probs +8probs Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Results on RTE3 data (each data set contains 800 problems)

  48. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion Outline • Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion

  49. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion What natural logic can’t do • Not a universal solution for textual inference • Many types of inference not amenable to natural logic • Paraphrase: Eve was let go = Eve lost her job • Verb/frame alternation: he drained the oil<the oil drained • Relation extraction: Aho, a trader at UBS…<Aho works for UBS • Common-sense reasoning: the sink overflowed<the floor got wet • etc. • Also, has a weaker proof theory than FOL • Can’t explain, e.g., de Morgan’s laws for quantifiers: Not all birds fly= Some birds don’t fly

  50. Introduction • Foundations of Natural Logic • The NatLog System • Experiments with FraCaS • Experiments with RTE • Conclusion :-) Thanks! Questions? What natural logic can do Natural logic enables precise reasoning about containment, exclusion, and implicativity, while sidestepping the difficulties of translating to FOL. The NatLog system successfully handles a broad range of such inferences, as demonstrated on the FraCaS test suite. Ultimately, open-domain textual inference is likely to require combining disparate reasoners, and a facility for natural logic is a good candidate to be a component of such a system. Future work: phrase-based alignment for textual inference

More Related