Weapon equipment question answering system based on BERT and knowledge graph

Abstract

Knowledge of weaponry and equipment is a crucial basis for formulating equipment utilization strategies and development pathways. To address issues such as data redundancy, high interaction difficulty, and low match accuracy of question answers, this paper constructs a Q&A system based on a knowledge graph for weaponry and equipment. The system achieves named entity recognition and classification of questions by fine-tuning the BERT model; it generates graph database query statements by filling named entities into question templates and generates answers by filling answer templates. Analysis of test results shows that this system is capable of effectively ranking correct answers at the top and has achieved a good balance between accuracy and comprehensiveness, although there is still room for improvement.

Keywords

• knowledge graph|bert model|weapon equipment