Abstract: Research of the resources in the thematic database of humanities and social sciences is valuable. However, the data utilization in the existing thematic databases has many defects and requires deep semantic processing. This paper studies the semantic architecture of the thematic database, as well as the key technologies. Then we conduct in-depth research on the entire process of the semantic processing, such as entity recognition, classification, hierarchical and non-hierarchical relationship extraction. Finally, this article uses the journal literature data in the humanities and social sciences thematic database as a case to construct the ontology, and uses Neo4j, the graph database for data storage, query and inference. The results show that the key technologies proposed in this paper can realize the deep semantic processing of the humanities and social sciences thematic database.

Key words: Ontology, Graph database, Relationship extraction, Entity recognition, Semantic retrieval