Resume Extraction and Validation from Public Information
Xiangyi Kong, Zhaoquan Gu, Le Wang, Lihua Yin, Shudong Li, Weihong Han · 2020
With the development of information technologies, knowledge graphs have attracted much attention from more and more fields, due to their usage in information retrieval, recommendation, etc. However, the scale of many knowledge graphs is relatively small due to lack of huge data, it is difficult to perform text understanding and knowledge reasoning on them. Consequently, it is necessary to combine different knowledge graphs. Nevertheless, in some scenarios, an attribute may have multiple values in different knowledge graphs, and some values may be wrong. Knowledge validation is a core and difficult point in knowledge graph fusion. In this paper, we study resume extraction and validation from many public resources. Specifically, we extracted personnel information from four encyclopedia websites; we also propose a new multi-truth discovery and validation method, which evaluates the credibility of attribute values through Internet public information. The proposed method can not only be applied to the research of knowledge graph fusion, but also can be widely adopted in various fields such as rumor discovery, text understanding, knowledge reasoning.