摘要：

Abstract:

【Purpose/significance】In the age of artificial intelligence, knowledge question answering system has become a typical de⁃

velopment trend of domain knowledge services. Constructing an academic domain knowledge graph is an indispensable condition and

knowledge foundation for achieving discipline knowledge Q&A system.【Method/process】For LIS academic knowledge graph construc⁃

tion problem, we used a semi-structured interview to investigate the existing platform and potential needs of LIS academic knowledge

Q&A and built a knowledge quiz demand model. According to the demands, we integrated CNKI, Web of Science, Baidu Academic,

Baidu Encyclopedia, and ScienceNet to obtain multi-source data as the core data sets. In addition, we explored the construction

scheme of knowledge graph, which uses the Python core library "selenium+beautifulsoup+pandas”, the co-occurrence analysis meth⁃

od for knowledge extraction, the Edit distance algorithm and the Jaccard algorithm for knowledge fusion and the Neo4j graph database

completed knowledge storage and achieved the visualization display.【Result/conclusion】Building a "Knowledge Extraction" academ⁃

ic knowledge graph as an example, we explored knowledge service functions such as semantic retrieval, knowledge question answer⁃

ing, and cooperative recommendation.【Innovation/limitation】This paper, from the perspective of multi-source data integration,

builds the LIS academic domain knowledge graph, through which the knowledge representation is more accurate and can effectively

meet the needs of users to accurately acquire domain knowledge. In the next research, the sample data will be expanded and the rela⁃

tionship between entities will be judged through logical reasoning.