This study aims to integrate structured information on the Web using crowdsourcing methods to build a robust knowledge base.Traditional pattern matching techniques have limitations in dealing with the incompleteness of web tables,espe-cially in discovering semantic correspondences between different columns.This article proposes a hybrid machine-crowdsourcing method to overcome the limitations of traditional pattern matching techniques.The method is based on understanding the semantics of tables,selecting the most valuable columns for crowdsourcing verification,and integrating the results to infer concepts of other columns to achieve table matching.This research provides an effective solution for web data integration and lays a solid foundation for knowledge base construction.
万方数据