Information Technology Reference
In-Depth Information
Identification of the specific relations in the proposed framework also allows us a
richer interpretation of anaphora which are represented by more than one word, that is,
compound nouns. In this case, the framework allows us to interpret the modifier-noun
with a relation, in addition to the head noun. As a simple example, the compound noun
battle fatigue , appearing after the clause “ The battle caused fatigue ” has a co-referential
relation to the noun fatigue , but in addition it also has some semantic relation (identified
later as CAUSE ) to the noun battle .
Hence, there are two novel aspects to this framework for interpreting anaphora.
Firstly it identifies a specific relation between the anaphor and its antecedent. Secondly,
it also interprets modifiers beyond using them to merely identify the antecedent for the
head noun, that is, it interprets them in the same way as the head noun. A consequential
effect of this is that an NP can have more then one antecedent. Thus this framework
enables us to determine the relational dependence of an anaphoric NP to all other NPs
in the discourse.
2
Related Works
NP anaphora resolution has received considerably less attention from computational
linguists compared to pronominal anaphora even though the proportion of NP anaphora
in natural discourses is either comparable to, or more then the proportion of pronominal
anaphora. The reason for this seems to stem from the fact that the problem of pronoun
resolution is much better defined compared to NP anaphora. This difference in com-
plexity of the problem also explains why whatever published work is available on NP
anaphora resolution, is predominantly focussed on NPs that are definite descriptions
(eg. [24,8,2,3]) with the accompanying task of identifying whether a definite NP is
anaphoric or not. NP resolution in these studies involves identifying a single previously
mentioned noun that the anaphoric NP refers to. Anaphora in these studies have been
studied as two categories; direct and associative . The direct category includes cases
in which an NP directly co-refers to another entity such as the case of he/John .The
associative category includes cases such as window/house . Some of the studies such
as [24] have gone a step further to specify the actual associative relation in terms of
synonymy 1 , hyponymy 2 and meronymy 3 . The motivation for these relations seems to
have risen from organization of the lexicon, WordNet [7] which is used to bridge the
meanings between the anaphor and the antecedent.
In this paper we propose a framework that presents an enhanced interpretation of
the generic bridging relations. The framework is based on recognizing that anaphora
is used in a way similar to another natural language phenomenon, namely compound
noun generation. A compound NP of the form noun + noun (N + N) consists of two
nouns which have some underlying semantic relations ([17,6,19]). According to these
studies, use of compound NPs is highly productive rather then lexical . In this productive
process, compound NPs are formed on the fly as a discourse is being produced, rather
then recalled and used from a lexicon. In this productive process, the semantic relation
1
Same meaning relation.
2
Same subset/superset relation.
3
Part/whole relation.
Search WWH ::




Custom Search