Method and system for improving question and answer accuracy of knowledge base

Abstract

The invention provides a method and a system for improving question and answer accuracy of a knowledge base. The method comprises the following steps: obtaining a user question to be answered, extracting a theme entity in the user question, retrieving the knowledge base by using the theme entity, taking path information of each obtained candidate answer as a candidate path, and preprocessing the user question to obtain vector representation of the user question; scoring each step relationship on the candidate path by utilizing the vector representation by utilizing an attention mechanism to obtain a relationship confidence coefficient of each step relationship on the candidate path, and summing all relationship confidence coefficients on the candidate path to obtain a path confidence coefficient of the relationship path; and sorting all the candidate paths according to the set path confidence coefficients, and outputting the candidate path with the highest path confidence coefficient as an answer result of the user question. According to the invention, the effect of the intermediate node in the whole relationship inference is enhanced, and the accuracy of the relationship inferenceis improved.

Description

technical field [0001] The present invention relates to the field of Internet technology and the field of relationship inference in big data analysis, and in particular to a knowledge base question answering method and system for relationship inference based on path information. Background technique [0002] The knowledge base question answering system is a research hotspot in the field of natural language processing. The user inputs a complete and colloquial question sentence, and the system can return a clear answer string by querying in the structured knowledge base. Knowledge in the knowledge base is usually stored in the form of triples, namely (head entity, relation, tail entity). Generally speaking, a knowledge base question answering system contains two core modules, namely, the subject entity inference module and the relation inference module. The topic entity inference is to find out the entities that the user is interested in in the user's question and link them ...

AI Technical Summary

Problems solved by technology

The method based on semantic analysis needs to learn how to construct structured query statements, and often requires a large amount of manually labeled data. This type of labeled data requires labelers to be familiar with the corresponding linguistic knowledge, which is expensive.

The embedding-based method ignores the relationship inference step, treats single-relationship questions and multi-relationship questions uniformly, regards all nodes within two hops connected to the subject entity in the knowledge base as candidate answers, and does not explicitly model multi-relationship questions The final impact of intermediate nodes on relationship inference. This method of encoding all nodes as encoded candidate answers is similar to information retrieval. When dealing with multi-relational problems, it does not make full use of the information in the question and knowledge base, making the result of relationship inference relatively vague

For example, in the typical multi-relational question "What is the height of Yao Ming's wife?", the traditional method first extracts all nodes within two hops connected to "Yao Ming" in the knowledge base, and finally calculates the encoded node and The similarity of the question sentences, but the question here is not the height of Yao Ming, nor the height of his parents or teammates, but the height of his wife Ye Li. The node returned by the traditional method is likely to be the height of Yao Ming or his related people. Not the height of his wife Ye Li

Images