Knowledge graph-based financial field knowledge question-answering method

Abstract

The invention discloses a knowledge graph-based financial domain knowledge question-answering method, which comprises the following steps: firstly, crawling financial data to establish a knowledge base, vectorizing entities through Word2vec, and calculating similarity among the entities by using a cosine similarity algorithm; based on a financial dictionary, adopting a maximum forward traversal method to traverse, cut and analyze questions raised by a user, then carrying out classification training on questions in the financial field through an RNN neural network, and carrying out data retrieval according to question categories; extracting words with higher similarity with the question keywords in the corpus by using a similarity algorithm to expand the vocabularies of the question sentences, and finally completing matching of database query sentences according to a template matching method; returning the queried result as an answer to the question to the user, and realizing visualization of the relationship between the financial entity attribute data and the entities through ECharts. According to the method disclosed in the invention, through the neural network training model andsimilarity comparison adaptive parameter adjustment, the accuracy of the question-answering system is effectively improved.

Description

technical field [0001] The invention belongs to the technical field of machine learning, and in particular relates to a knowledge graph-based question answering method in the financial field. Background technique [0002] The knowledge graph is a graph-structured database. The graph is represented by entities as nodes and relationships as edges. It has important applications in searching, analyzing and recommending data. With the rapid development of knowledge map technology, a single knowledge map cannot meet the actual needs. Therefore, the multi-source knowledge embedding technology of knowledge map is gradually mature. When completing the knowledge extraction step, it is necessary to use natural language processing technology to extract entities related to financial topics, and store the corresponding relationship in a relational database. A variety of convolutional neural network models have been proposed to fit various tasks in natural language processing. Completing...

AI Technical Summary

Problems solved by technology

[0016] In the technical field of natural language processing, although the knowledge map can basically realize the question-and-answer function, it cannot realize the precise search function according to the specific semantics, and the hidden connection between entities cannot be effectively mined

Images