<?xml version="1.0" standalone="yes"?>
<Paper uid="P95-1017">
  <Title>Evaluating Automated and Manual Acquisition of Anaphora Resolution Strategies</Title>
  <Section position="8" start_page="295" end_page="295" type="relat">
    <SectionTitle>
5 Related Work
</SectionTitle>
    <Paragraph position="0"> Anaphora resolution systems for English texts based on various machine learning algorithms, including a decision tree algorithm, are reported in Connolly et al. (Connolly et al., 1994). Our approach is different from theirs in that their decision tree identifies which of the two possible antecedents for a given anaphor is &amp;quot;better&amp;quot;. The assumption seems to be that the closest antecedent is the &amp;quot;correct&amp;quot; antecedent. However, they note a problem with their decision tree in that it is not guaranteed to return consistent classifications given that the &amp;quot;preference&amp;quot; relationship between two possible antecedents is not transitive.</Paragraph>
    <Paragraph position="1"> Soderland and Lehnert's machine learning-based information extraction system (Soderland and Lehnert, 1994) is used specifically for filling particular templates from text input. Although a part of its task is to merge multiple referents when they corefer (i.e. anaphora resolution), it is hard to evaluate how their anaphora resolution capability compares with ours, since it is not a separate module. The only evaluation result provided is their extraction result.</Paragraph>
    <Paragraph position="2"> Our anaphora resolution system is modular, and can be used for other NLP-based applications such as machine translation. Soderland and Lehnert's approach relies on a large set of filled templates used for training. Domain-specific features from those templates are employed for the learning. Consequently, the learned classifiers are very domain-specific, and thus the approach relies on the availability of new filled template sets for porting to other domains.</Paragraph>
    <Paragraph position="3"> While some such template sets exist, such as those assembled for the Message Understanding Conferences, collecting such large amounts of training data for each new domain may be impractical.</Paragraph>
    <Paragraph position="4"> Zero pronoun resolution for machine translation reported by Nakaiwa and Ikehara (Nakaiwa and Ikehara, 1992) used only semantic attributes of verbs in a restricted domain. The small test results (102 sentences from 29 articles) had high success rate of 93%. However, the input was only the first paragraphs of newspaper articles which contained relatively short sentences. Our anaphora resolution systems reported here have the advantages of domain-independence and full-text handling without the need for creating an extensive domain knowledge base.</Paragraph>
    <Paragraph position="5"> Various theories of Japanese zero pronouns have been proposed by computational linguists, for example, Kameyama (Kameyama, 1988) and Walker et aL (Walker et al., 1994). Although these theories are based on dialogue examples rather than texts, &amp;quot;features&amp;quot; used by these theories and those by the decision trees overlap interestingly. For example, Walker et ai. proposes the following ranking scheme to select antecedents of zero pronouns.</Paragraph>
    <Paragraph position="6">  In examining decision trees produced with anaphoric type identification turned on, the following features were used for QZPRO-ORG in this order: topicalization, distance between an anaphor and an antecedent, semantic class of an anaphor and an antecedent, and subject NP. We plan to analyze further the features which the decision tree has used for zero pronouns and compare them with these theories.</Paragraph>
  </Section>
class="xml-element"></Paper>