基于知识图谱的广西文化旅游问答系统研究与实现

Research and Implementation of Guangxi Cultural Tourism Question Answering System based on Knowledge Graph

关键词:  知识图谱  问答系统  深度学习  自然语言处理  命名实体识别  关系抽取

Key words:  knowledge graph  question answering system  deep learning  natural language processing  named entity recognition  relationship extraction

摘要:

当前的旅游咨询服务还只是为用户提供自主网络搜索返回的碎片化信息，尚未能将地方特色文化智能反馈给用户。针对此实际情况，本研究基于广西民族文化旅游知识图谱，对广西民族文化旅游问答系统的关键技术加以研究，并设计相应的问答系统，在解决实际需求的同时，尝试提高用户咨询体验满意度。根据问答系统（Question Answering System，QA）结构，本研究设计并实现了基于BERT的命名实体识别模块（BERT based Entity_identification Model，BEiM），基于模版的关系抽取模块（Template based Relationship_extraction Module，TReM）和基于知识图谱的匹配推理模块（Knowledge Graph based Matching Module，KGMM）。在上述关键技术基础上，实现了广西文化旅游问答系统，并给出相关实验测试和应用效果。本研究构建的知识问答系统能够帮助游客高效地找到当地旅游的相关知识，提高游客自助服务的效率。对于人工智能助力广西旅游业的发展而言，本研究无疑是一项具有重要意义的工作。

Abstract:

The current tourism consulting service is only to provide users with fragmented information returned by independent web search,and has not yet been able to feed back the local characteristic culture intelligently to users. In response to this actual situation,the key technologies of the Guangxi ethnic culture and tourism question answering system based on the knowledge map of Guangxi ethnic culture and tourism are studied in this article. And the corresponding question and answer system is designed to try to improve the satisfaction of users consultation experience while solving the actual needs. According to the structure of the question answering system,this article designs and implements a BERT-based named entity identification module (BERT based Entity_identification Model,BEiM),a template-based relationship extraction module (Template based Relationship_extraction Module,TReM) and a knowledge graph-based matching inference module (Knowledge Graph based Matching Module,KGMM).On the basis of the above key technologies,the Guangxi cultural tourism question answering system is implemented,and the relevant experiments and application effects are given. The knowledge question answering system constructed in this research can help tourists find the relevant knowledge of local tourism efficiently and improve the efficiency of tourists' self-service. This research is undoubtedly an important work for artificial intelligence to help the development of Guangxi tourism.