Joint Entity and Relation Extraction Method Based on Knowledge Representation Attention

Relation extraction is a fundamental task in natural language processing and is a key step in information extraction tasks and construction of large-scale knowledge graphs, etc. Knowledge graph ontology information is useful for guiding triplet construction, but existing methods do not make full use of relevant information such as relation. Therefore, this paper proposes a joint extraction method of subject-aware entity relation combined with knowledge relation representation. The relation information of knowledge graph is extracted using TransE. Firstly, the subject entity information is extracted based on the text representation, the subject entity information and the relation representation are computed with attention to enhance the connection between the subject and the relation, and the subject information combined with the relation information is used as input to extract the triplets corresponding to the current subject. Compared with the baseline model, the method achieves an increase in F1 value on the Baidu DuIE test set. The experimental results show some improvement in the accuracy and recall of the triplets, and the attention based on relation representation enhances the dependencies between entity pairs and relations. The effectiveness of relation representation modeling is verified.