This paper proposes a method to analyze Japanese anaphora, in which zeropronouns (omitted obligatory cases) are used to refer to preceding entities(antecedents). Unlike the case of general coreference resolution, zero pronounshave to be detected prior to resolution because they are not expressed indiscourse. Our method integrates two probability parameters to perform zeropronoun detection and resolution in a single framework. The first parameterquantifies the degree to which a given case is a zero pronoun. The secondparameter quantifies the degree to which a given entity is the antecedent for adetected zero pronoun. To compute these parameters efficiently, we use corporawith/without annotations of anaphoric relations. We show the effectiveness ofour method by way of experiments.
展开▼