Research on Question Matching of Operation System FAQ with Knowledge Graph and Image Information
Conference: CIBDA 2022 - 3rd International Conference on Computer Information and Big Data Applications
03/25/2022 - 03/27/2022 at Wuhan, China
Proceedings: CIBDA 2022
Pages: 7Language: englishTyp: PDF
Authors:
Zhao, Haomin; Liu, Yu; Gu, Jinguang (College of Computer Science and Technology, Wuhan University of Science and Technology, Wuhan Hubei, China & Key Laboralory of Intelligent Information Processing and Real-time Industrial System in Hubei Province Wuhan Hubei, China & Institute of Big Data Science and Engineering, Wuhan University of Science and Technology, Wuhan Hubei, China)
Abstract:
The FAQ system is to match the questions asked by the user with the questions in the FAQ library, and subsequently return the answers to the questions according to the smenatic similarity between questions. The existing matching methods of question pair are not specially designated for the OS domain and do not incorporate domain knowledge. The screenshots of Operating System provided by user usually contain important information, which can be uesed to improve the effectiveness of question matching method. In this paper, we evaluate the important of text in image with the highlighted evaluation, location evaluation, and text evaluation, and filter out important text for question matching. In order to use knowledge of Operation System, we introduce a pretraining model incorporating the domain knowledge that can be used for problem matching. In the experiment, the method with knowledge graph and image information can fully improve the accuracy rate, recall rate and F1 score of the matching reuslts compared to the mainstream method.