Design And Implementation Of A Government Knowledge Graph Question-answering System Based On Deep Learnin

In recent years,with the rapid development of computer technologies such as artificial intelligence and big data,it has become a trend to use these emerging technologies to improve the intelligence level of government governance and empower of a digital government.As an important part of the construction of digital government,the government question answering system is of great value to improve the efficiency of government administration.However,the existing government question answering systems mostly rely on artificial question templates,and match data information according to keywords to return to users,which makes the question answering system has the universality of the lower,and it is hard to correctly understand the semantics of complex problems,resulting in limited quality of reply content and poor user experience.To solve the above issues,this thesis constructs the knowledge as the government knowledge base,studies the question answering method based on deep learning,and designs and implements an intelligent question answering system for government affairs.The main research contents and innovations of this thesis are as follows:(1)The construction of knowledge graph of government affairs.To address this problem that the lack of public knowledge graph data in the field of government service.Firstly,this thesis implements the crawler system based on the website crawler tools of Beautiful Soup to obtain the data of public government service websites,then designs the pattern layer of government knowledge graph based on the extracted data,and uses Neo4 j graph database to complete the visualization of government knowledge graph data layer.(2)Question answering method in government affairs based on deep learning.To solve the problem of NER,a BERT-Bi LSTM-MHA-CRF method of NER based on pretraining model and multi-head self-attention mechanism is studied in this thesis.The model extracts the deep semantic information of question better by introducing the knowledge acquired by large-scale text pretraining combined with the weight analysis of attention mechanism.To solve the problem of IR,a BERT-RCNN model with multilayer hidden layer parameters is used to complete IR.Experiments show that the model proposed in this thesis has achieved good results.(3)Design and implementation of question answering system for government affairs knowledge graph.Through the demand analysis of question answering system for government affairs,the knowledge graph of government affairs is used as the knowledge base,and the question-answering method based on deep learning is used as the core to design and implement the question answering system in the field of government affairs.The functional test and non-functional test of the system are carried out,and the test results show that the system can meet the design requirements.The knowledge graph is introduced into the question answering system,which can provide rich background knowledge for semantic understanding.The combination of deep learning model can more in-depth identify the question semantics,so as to promote the intelligent development of government question answering field,which is of great significance to realize the efficient public service.