Digital Approaches to Name Disambiguation of Chinese Historical Figures

When integrating biographical data extracted from 2,000+ local gazetteers into the China Biographical Database (CBDB), we need to identify and link records of the same person--the act of “disambiguating” them. Traditional Chinese naming customs pose big challenges to this, especially for the gazetteer dataset containing 0.12 million records and 90k unique names of imperial government officials. Also, useful variables are missing in numerous entries in these gazetteers. My presentation analyzes solutions to disambiguating identical personal names in Chinese script. First, we identified the individuals who repeatedly took official posts in the same locality. Then, we cross-tabulated the overlap of content in multiple gazetteers. Finally, we corroborated the remaining data with external datasets e.g. CGED-Q of the Lee-Campbell research group. Through doing so we have disambiguated 51k personal names with optimal precision. Such task is only possible if done digitally. The techniques explored in this study will also be useful for disambiguation and Named Entity Recognition of other large-scale unstructured data in non-Latin script.