Research And Implementation Of Financial Question Answering System Based On Knowledge Graph

With the increasing per capita GDP,a large number of people are involved in the financial market,but geting the needed information fast and accurately is a difficult thing.Because for traditional search engines,after the user enters a question,through shallow semantic matching,the various web pages returned not only cannot understand the user’s true intention,but also requires the user to spend time refining the answer.After the user asks a question,the question answering system will locate the user’s real thoughts and return intuitive and refined answers,thereby improving the user experience.This paper combines financial information with natural language processing technology,designs a financial question answering algorithm based on knowledge graph,and completes a financial question answering system based on knowledge graph.The main work is as follows:(1)Build a financial knowledge graph in a bottom-up manner.First of all,we need to obtain data,obtain data through interface calls and design crawler programs,and clean the data.Then,according to certain rules,the acquired data is extracted from entities,attributes and relationships,and converted into triples.Finally,the obtained triple information is stored in the neo4 j graph database to realize visual query.(2)Algorithm design of financial question answering system based on knowledge graph.The core of the algorithm is to link the keywords in the questions asked by users to the knowledge graph,and return intuitive and concise answers.This process mainly includes three tasks,entity recognition,classification and answer query of financial questions.Use the Bi LSTM-CRF named entity recognition model to identify the entities and attributes of the question,and use the naive Bayes classifier to classify the question to obtain a query template.query and return the answer.At the same time,named entity recognition experiments and question classification experiments were set up.On the constructed dataset,Bi LSTM-CRF achieved 0.9 F1 value,and Naive Bayes classifier achieved 0.95 F1 value.(3)Construction of financial question and answer system.To build a financial question answer system based on web,this paper uses the Flask framework.The whole system is grouped into three modules,data module,question and answer module,and front-end display module.To sum up,a financial question answering system based on knowledge graph has been successfully built by this paper,and the function and performance of the system have also been tested.The accuracy rate of question answering reaches 82.5%.The results of test exhibited that the financial question answering system designed in this paper can help users quickly get information about financial products.