Comparing Knowledge Sources for Nominal Anaphora Resolution

Abstract

We compare two ways of obtaining lexical knowledge for antecedent selection in other-anaphora and definite noun phrase coreference. Specifically, we compare an algorithm that relies on links encoded in the manually created lexical hierarchy WordNet and an algorithm that mines corpora by means of shallow lexico-semantic patterns. As corpora we use the British National Corpus (BNC), as well as the Web, which has not been previously used for this task. Our results show that (a) the knowledge encoded in WordNet is often insufficient, especially for anaphorantecedent relations that exploit subjective or context-dependent knowledge; (b) for otheranaphora, the Web-based method outperforms the WordNet-based method; (c) for definite NP coreference, the Web-based method yields results comparable to those obtained using WordNet over the whole data set and outperforms the WordNet-based method on subsets of the data set; (d) in both case studies, the BNC-based method is worse than the other methods because of data sparseness. Thus, in our studies, the Web-based method alleviated the lexical knowledge gap often encountered in anaphora resolution and handled examples with context-dependent relations between anaphor and antecedent. Because it is inexpensive and needs no hand-modeling of lexical knowledge, it is a promising knowledge source to integrate into anaphora resolution systems.

Towards the self-annotating web

“…In fact, several researchers have shown that using the Web as a corpus is an effective way of addressing the typical data sparseness problem one encounters when working with corpora (compare [14], [21], [23], [25]). Actually, we subscribe to the principal idea by Markert et al [23] of exploiting the Google TM API.…”

“…• Others use the Web to cope with data sparseness problems in tasks that require statistics about possible semantic relationships ( [1,14,21,23]). …”

Analysis and Reference Resolution of Bridge Anaphora across Different Text Genres

“…Bridging anaphora have been extensively investigated by Poesio et al [17,16,25,15,10]. Both the typology of bridging references and di↵erent information sources to improve automatic resolution were studied.…”

“…Another method is to use syntactic patterns that are able to express partwhole relations such as "the x of the y". If such patterns of a potential antecedent and anaphor are found with a high frequency in a corpus [14] or on the web [15,10], this provides a strong clue that they could stand in a bridge relation. Besides the above-mentioned features, in [15] features expressing salience, such as utterance distance and first-mention were also studied.…”

“…Based on this analysis and previous studies on resolving bridge relations (e.g. [16,15,10]), we also present a first version of an automatic resolution system for Dutch bridge relations. We wish to investigate to what extent a standard mention-pair coreference resolver is applicable to the task, how it behaves across di↵erent genres, and study the usefulness of WordNet (WN) for resolving bridge relations in our data sets since computing semantic distance based on WN has proven a useful feature in previous work [15].…”