Research And Implementation Of Q&A System Based On Chinese Knowledge Graph In Financial Field

With the rapid development of the Internet,the question answering system is favored to provide intelligent knowledge service for users.From the perspective of the financial field,this thesis sorted out and structured the Chinese data in the financial field,constructed the knowledge graph based on the uncertainty and diversity of the question asked by users,researched and realized the question answering system(CF-KGQA)in the financial field based on the Chinese knowledge graph.The main work is as follows:1.Obtain and store data and build a knowledge map with characteristics of the financial field.(1)Set up a distributed crawler system with one master and ten slaves,and in order to ensure the security of data storage,set up a database cluster that can be masterslave backup.(2)Define the concepts of entities and relationships between entities in the knowledge graph.When constructing the knowledge graph,the structure of the graph should not only consider the characteristics of the financial field,but also be adjusted according to the actual needs of the question answering system.2.Propose a deep learning-based analysis method for interrogative semantics in the financial field.Present a semantic Dependency Graph Parsing(SDGP),Bidirectional Long Short-Term Memory(BLSTM)and Conditional Random Field(CRF)based on IFLYTEK open platform.(1)Combine the BLSTM and CRF's Named Entity Recognition(NER)algorithm to perform named entity recognition on the question and obtain a sequence containing character label information.(2)Use the open platform of IFLYTEK,Web API,to perform semantic dependency graph analysis on the question,obtain a sentence representation containing semantic dependency information,and then combine the named entity recognition results in(1)to obtain a more accurate semantic dependency graph through dependency reduction.The experimental results show that the accuracy rate,recall rate and F1 value of the proposed method compared with the semantic dependency analysis effect of Language Technology Platform(LTP)on the self-built dataset of about 140,000 question questions in the financial field.Increased by 33.4%,33.9%,and 34.2%,respectively,this method can effectively perform semantic dependency analysis on questions in the financial field.3.Design and implement a question answering system based on knowledge graph.Based on previous theories and experiments,the question answering system applies the knowledge graph and deep learning-based financial domain question semantic dependency analysis methods.The knowledge graph module,front-end display module,and question answering module are designed as three functional modules.Web pages are displayed in the form of dynamic graphics.