[Objective]Based on the perspective of knowledge element, this paper explores the extraction and integration of domain knowledge from heterogeneous data to enrich the semantic information of knowledge representation. [Methods]The existing knowledge element semantic description model was optimized, and the knowledge extraction and representation method based on the knowledge element semantic description model was proposed, and the application was carried out in the information retrieval field.[Results]4200 knowledge elements and 3020 entities in information retrieval field were extracted from Wikipedia and two classic textbooks, which could be used to query the relationship between knowledge elements and their entities. [Limitations]The semantic relations among knowledge elements are not fully mined, and the process of knowledge extraction is not fully automatic. [Conclusions]To a certain extent, this paper breaks through the limitation that knowledge representation semantics is not rich in existing research, and provides a new perspective for domain knowledge service.