Domain question and answer-oriented knowledge graph construction method

Abstract

The invention provides a domain question and answer-oriented knowledge graph construction method, which comprises the following steps: S1, collecting common questions and answers of a user to a domainon the basis of the professional domain, classifying the questions to form a question set, and training a Chinese pre-training classification model on the basis of the data; s2, performing word segmentation and part-of-speech tagging classification on each question in the question set by using the question set in the step S1, constructing a knowledge graph of the current classification by using aword dictionary obtained after each question formed by the same category is subjected to word segmentation and part-of-speech tagging, and storing the knowledge graph in a graph database; s3, performing sub-category division on the user problems according to actual requirements, performing retrieval on retrieval rules used in the knowledge graph of the corresponding category in the graph databaseaccording to sub-category results and the keyword set, and obtaining retrieval labels; and S4, retrieving corresponding labels, extracting questions represented by the labels in the database, and returning answers corresponding to the questions to the user.

Description

technical field [0001] The invention relates to the fields of natural language processing, knowledge map construction, and professional field question answering, and in particular to a method for building knowledge map oriented to field question answering. Background technique [0002] The main purpose of domain-based question answering systems is different from general-purpose question-answering systems. It is mainly to answer professional questions in related professional fields for users. Usually, professional fields involve many proprietary words, and conventional question-answering methods cannot understand the semantic associations of these proprietary words. At the same time, the traditional question-and-answer method has insufficient relevance of proprietary words, resulting in inaccurate answers and other problems. [0003] Therefore, by analyzing the collected questions and answers frequently asked by users in the field through natural language processing methods, ...

AI Technical Summary

Problems solved by technology

[0007] In view of the fact that the professional words of the existing domain-type question-and-answer are difficult to semanticize and the accuracy of the answer is low, the present invention mainly uses the strong correlation between the nodes of the knowledge map to formulate different categories according to the requirements, and based on the words extracted from the question data set. Construct the domain Q&A knowledge map; if a user asks a question, use the domain Q&A knowledge map to find the best answer through the designed special reasoning rules and return it to the user

Images