0

I am going through the Lappin and Leass "An algorithm of pronominal anapora resolution". I am particularly confused about some of the conditions for the syntactic filter.

Background: (Section 2.1.1 of the above paper)

The agreement features of an NP are its number, person, and gender features. We will say that a phrase P is in the argument domain of a phrase N iff P and N are both arguments of the same head.

We will say that P is in the adjunct domain of N iff N is an argument of a head H, P is the object of a preposition PREP, and PREP is an adjunct of H.

P is in the NP domain of N iff N is the determiner of a noun Q and (i) P is an argument of Q, or (ii) P is the object of a preposition PREP and PREP is an adjunct of Q.

A phrase P is contained in a phrase Q iff (i) P is either an argument or an adjunct of Q, i.e., P is immediately contained in Q, or (ii) P is immediately contained in some phrase R, and R is contained in Q.

Then the go on to write 6 conditions for this purpose. These are the Conditions on NP-pronoun non-coreference. I fail to understand that given a constituency tree how to apply all of this? Could anyone explain the examples to me? I strongly feel I have flawed understanding that is resulting in my code to wrongly process sentences.

  1. P is in the argument domain of N.

Sentence: "She likes her" & "John seems to want to see him" How does one differentiate between co referential valid sentences from the above?? The second sentence stumps me the most.

(S (NP she)
   (VP likes
       (NP her))

(S (NP John)
   (VP seems
       (S (VP to
              (VP want
                  (S (VP to
                         (VP see
                             (NP him)))))))))
  1. P is in the adjunct domain of N.

Sentence: "She sat near her" Here "her" is supposedly adjunct of Head H. How do I know this based on the tree?

(S (NP she)
   (VP sat
       (PP near
           (NP her))))
  1. P is an argument of a head H, N is not a pronoun, and N is contained in H.

Sentence1: He believes that the main is amusing.

Sentence2: This is the man he said John wrote about.

(S (NP He)
   (VP believes
       (SBAR that
             (S (NP the man)
                (VP is
                    (NP amusing))))))

(S (NP This)
   (VP is
       (NP (NP the man)
           (SBAR (S (NP he)
                    (VP said
                        (SBAR (S (NP John)
                                 (VP wrote
                                     (PP about))))))))))
  1. P is in the NP domain of N. Sentence: John's portrait of him is interesting
(S (NP (NP (NP ohn 's)
       portrait)
   (PP of
       (NP him)))

(VP is (ADJP interesting)))

Thank You.

1 Answers1

1

The general idea is that pronouns cannot corefer in the same S or NP headed tree. One term for this is binding in the context of C-command, not sure what current theories call it. The wikipedia page is somewhat helpful https://en.wikipedia.org/wiki/C-command. In such situations the way to refer pronominally is to use a reflexive, so "John loves him" becomes "John loves himself."

All of Lappin and Lease's rules map fairly standard notions of C-command and pronoun coreference restrictions onto standard bracketed parse trees. Note that these parse trees need an additional step to determine what NPs are arguments to the same predicates/modifiers. That may be confusing you. So (S (NP John) (VP loves (NP him))) needs to be mapped to loves(John,him). The general idea is that coref restritions apply to arguments to predicates. For example "John loves the man who hates him" has no coref restriction between "John" and "him". Note that "the man" cannot corefer with "him."

If you want to make your life easy just restrict coreference within S and NPs that share arguments. Not perfect but maybe close enough.

Breck

Breck Baldwin
  • 151
  • 4