I. Introduction
Knowledge graphs establish connections between entities and concepts in a structured organizational form [1]. Knowledge graphs provide favorable data support for natural language processing tasks, such as intelligent semantic question answering and automatic recommendation [2]. Constructing a knowledge graph requires mining noise-less, high-quality data from a large number of information sources, including structured data, semistructured data and unstructured data. There are uneven quality, different semantic structures, and inconsistent representations in these data. How to effectively organize and compute the information obtained from different data sources is the technical challenge of knowledge graph construction.