Research on Toponym Resolution in Chinese Text

This paper explores approaches for Toponym resolution in Chinese text,and proposes a geo-parsing approach based on conditional random fields and discourse toponym relations,and a geo-coding approach based on partial fuzzy matching and cognitive salience calculation.The proposed geo-parsing approach deals with the recognition of toponym in three major steps.The experiment shows that the key factors that may influence the performance of toponym resolution in Chinese text are the coverage of gazetteer,the performance of geo-parsing and the performance of semantic disambiguation of toponyms.In our experiment,there are about 17% toponyms can not locate their semantics in the gazetteer.Ambiguity in geo-parsing and geo-coding are the next prominent factors that affect the overall performance of toponym resolution.