1. Reference Resolution

2. Sue bought a cup of coffee and a donut from Jane. She met John as she left. He looked at her enviously as she drank the coffee. It was delicious.

5. How could a computer resolve these references? Complete model of dialogue too complicated Are there any simpler ways?

6. Hard Constraints on Coreference Number agreement Person and case Gender Agreement Syntactic Agreement Selectional Restrictions

7. Number agreement John and Mary loaned Sue a cup of coffee. Little did they know the magnitude of her addiction. SingularPluralUnspecified She,her, he, him, his, it We, us, they, them you

8. Person and Case Agreement FirstSecondThird NominativeI,weyouhe,she,they Accusativeme,usyouHim,her,them Genitivemy,ouryourHis,her, their

9. Gender Agreement *John has a coffee machine. She loves it.

10. Syntactic Agreement Reflexives (himself, herself…) have strong constraints on what syntactic positions they can appear in John bought himself a cup of coffee. *John bought him a cup of coffee.

11. Selectional Constraints Jim bought a coffee from the store. He drank it quickly.

12. Also : Preferences Recency Grammatical Role Repeated Mention Parallelism Verb Semantics  Based on Salience

13. Recency John had a pop-tart. Bill had a jelly donut. Mary wanted it. Recent Entities are more salient

14. Grammatical Role “Sue bought a cup of coffee and a donut from Jane. She met John as she left.” Entities in subject position are more salient

15. Repeated Mention John went to the store to buy coffee. He loves coffee. He drinks 5 cups a day. At the store, Bill sold him a cup. He was delighed. Entities mentioned more frequently are more salient

16. Parallelism John bought coffee from Jim in the morning. Sue bought coffee from him in the evening. Even with preferences to the contrary (grammatical role) the syntactic parallelism strongly prefers [him = Jim]

17. Verb Semantics John telephoned Bill. He was jonesing for coffee. John criticized Bill. He was jonesing for coffee. Perhaps salience of different elements in the sentence changes with respect to the verb used.

18. Algorithms --- How to integrate these preferences? Constraints are easy to use : reject all hypothesis which violate the hard constraints (if you can accurately detect the constraints!) Preferences more difficult – how can one integrate these different preferences?

19. Lappin and Leass Use Weighting Scheme Weight each mention, cut values in half at each new sentence PreferenceWeight Sentence recency100 Subject emphasis80 Existential emphasis70 Accusitive (direct object) emphasis80 Indirect object, oblique compliment40 Non-adverbial emphasis50 Head noun emphasis80

20. Lappin & Leass in action Sue read a book on peanuts at the library. Recency : 100 Object : 50 Head Noun : 80 Non-Adverbial: 50 Recency : 100 Object : 50 Non-Adverbial: 50 Recency : 100 Head Noun : 80 Non-Adverbial: 50

21. Hobbs Tree Search Algorithm Given parse trees, search them in a specific order to find the most likely referent

22. Hobbs in Detail 1.Begin at NP 2.Go up tree to first NP or S. Call this X, and the path p. 3.Traverse all branches below X to the left of p. Propose as antecedent any NP that has a NP or S between it and X 4.If X is the highest S in the sentence, traverse the parse trees of the previous sentences in the order of recency. Traverse left-to-right, breadth first. When a NP is encountered, propose as antecedent. If not the highest node, go to step 5.

23. Hobbs cont. 5.From node X, go up the tree to the first NP or S. Call it X, and the path p. 6.If X is an NP and the path to X did not pass through the nominal that X dominates, propose X as antecedent 7.Traverse all branches below X to the right of the path, in a left-to-right, breadth first manner. Propose any NP encountered as the antecdent 8.If X is an S node, traverse all brnaches of X to the right of the path but do not go below any NP or S encountered. Propose any NP as the antecedent.

24. Charniak, Ng and Hale Statistical integration of preferences Similar to Lappin Leass –Uses distance (recency constraint) [d], syntactic position (gramatical role emphasis) [s], gender number and animacy [W], verb selectional constraints [l,h], and mention (repeated mention counts) [M]

25. CNH Model P( a(p) ) = a | p, h, W, t, l, s, d, M) = P(a|M)P(d h | a) P(W | h,t,l,a) P(p|w a ) –P(a|m a ) = # of mentions –P(d h | a) = distance as measured by Hobbs algorithm –P(W|h,t,l,a) = judges emphasis in the sentence –P(p| w a ) = integrates number, animacy, gender