Automatic Question Answering Technology of Attention Mechanism Based on Knowledge Graph

Abstract

The automatic answer technology of search engine and the automatic answer system based on intelligent machine lack data cleaning steps, and there is a problem that the automatic question answering effect is poor due to the low matching correct answer rate. A research on automatic question answering technology based on Attention mechanism of knowledge graph is proposed. Stores and manages knowledge graphs through Jena. It goes through three processes of ontology data cleaning, entity graphing and linking, and data type labeling, and integrates the knowledge graph oriented to question answering system. The Attention mechanism is used to retrieve automatic question-and-answer keywords, continuously adjust parameters, analyze the relationship between words, and design the automatic question-and-answer process. The experimental analysis is carried out in PyCharm environment, and the matching correct answer rate and the matching result evaluation mechanism are taken as the experimental evaluation indexes. From the experimental results, it can be seen that the matching correct answer rate of this technology can reach up to 93%, which can highlight the user preference and supplement the problem expansion research.