Automatic Pronominal Anaphora Resolution in English Texts

1 Automatic Pronominal Anaphora Resolution in English Texts Tyne Liang and Dian-Song Wu Department of Computer and Information Science National Chiao Tung University Hsinchu, Taiwan Email: tliang@cis.nctu.edu.tw; gis90507@cis.nctu.edu.tw Abstract Anaphora is a common phenomenon in discourses as well as an important research issue in the applications of natural language processing. In this paper, the anaphora resolution is achieved by employing WordNet ontology and heuristic rules. The proposed system identifies both intra-sentential and inter-sentential antecedents of anaphors. Information about animacy is obtained by analyzing the hierarchical relation of nouns and verbs in the surrounding context. The identification of animacy entities and pleonastic-it usage in English discourses are employed to promote the resolution accuracy. 1. Introduction 1.1 Problem description Anaphora resolution is vital for areas such as machine translation, summarization, question-answering system and so on. In machine translating, anaphora must be resolved for languages that mark the gender of pronouns. One main drawback with most current machine translation systems is that the translation usually does not go beyond sentence level, and so does not deal with discourse understanding successfully. Inter-sentential anaphora resolution would thus be a great assistance to the development of machine translation systems. On the other hand, many of automatic text summarization systems apply a scoring mechanism to identify the most salient sentences. However, the task result is not always guaranteed to be coherent with each other. It could lead to errors if the selected sentence contains anaphoric expressions. To improve the accuracy of extracting important sentences, it is essential to solve the problem of anaphoric references in advance. 2 Pronominal anaphora is the most common phenomenon which the pronouns are substituted with previous mentioned entities. This type of anaphora can be further divided into four subclasses, namely, Nominative: {he, she, it, they} Reflexive: {himself, herself, itself, themselves} Possessive: {his, her, its, their} Objective: {him, her, it, them} However, the usage of “it” can also be a non-anaphoric expression which does not refer to any items mentioned before and is called expletive or pleonastic-it [Lappin and Leass, 94]. Although pleonastic pronouns are not considered anaphoric since they do not have an antecedent to refer to, yet recognizing such occurrences is essential during anaphora resolution. In [Mitkov, 01], the non-anaphoric pronouns are in average of 14.2% from a corpus of 28,272 words. Definite noun phrase anaphora occurs in the situation that the antecedent is referred by a general concept entity. The general concept entity can be a semantically close phrase such as synonyms or superordinates of the antecedent [Mitkov, 99]. The word one has a number of different uses apart from counting. One of the important functions is as an anaphoric form. For example: Intra-sentential anaphora means that the anaphor and the corresponding antecedent occur in the same sentence. Inter-sentential anaphora is where the antecedent occurs in a sentence prior to the sentence with the anaphor. In [Lappin and Leass, 94], there are 15.9% of Inter-sentential cases and 84.1% Intra-sentential cases in their testing result. In the report of [Mitkov, 01], …

Content truncated to 500 words