Research on Intelligent Construction Method of the TRIZ Analysis Matrix

Subjects: Library Science，Information Science >> Information Science submitted time 2022-06-09

Xing Sisi Qian Li

Abstract：

[Objective] Based on deep learning and domain knowledge graph technology, study the automatic construction method of TRIZ analysis matrix to provide knowledge support for technological innovation. [Methods] First, facing the domain requirements, the general "technical elements" are extended and refined into entity and relationship categories, and the pattern layer design of domain knowledge graph is completed;Then, based on the BERT pre-trained language model, research and design an intelligent tool to automatically identify knowledge entities and relationships from patent documents, and realize the automatic extraction of entity relationships;Finally, the graph database is used to complete the construction of the domain knowledge graph, and the knowledge entities and relationships are required to read based on the knowledge query, so as to realize the automatic construction of the TRIZ analysis matrix. [Results] After empirical verification, the F1 score of the BERT-MH+CRF entity recognition model for patents in the field of thin-film magnetic head technology constructed in this paper is 84.93%, and the F1 score of the BERT-MH+softmax relation extraction model is 63.7%. [Limitations] The lack of application of knowledge reasoning technology cannot reveal the potential knowledge associations between entities, and the quality of the constructed TRIZ analysis matrix is still insufficient. [Conclusions] The proposed method can achieve the goal of automatic construction of TRIZ analysis matrix, which can provide effective knowledge support for technological innovation.

Peer Review Status:Awaiting Review

 Hits 3968  Downloads 721  Comment 0