Intelligent question answering method for search engine knowledge base

Abstract

A search engine knowledge base-oriented intelligent question and answer method comprises the steps of constructing a reasoning path of symbol Markov properties through dynamic search path reasoning and deep reinforcement learning on the basis of a search engine knowledge base, realizing path information coding and action space probability distribution calculation on the basis of an LSTM and a feedforward neural network, and setting a judgment condition according to vector representation of the search reasoning path; based on the judgment condition and the reasoning path during question answer searching, achieving the intelligent question answering of the search engine based on the knowledge base. The method does not need to define rules, does not limit the length of a reasoning path, can be used for a complex question and answer reasoning process based on a search engine knowledge base, and realizes efficient and accurate search engine intelligent question and answer.

Description

technical field [0001] The invention relates to the field of intelligent question answering of search engines, in particular to an intelligent question answering method for search engine knowledge bases. Background technique [0002] The intelligent question answering of the search engine, that is, given a natural language question to the search engine, the corresponding entity is searched from the existing knowledge base as the answer to the question. Specifically, it involves entity recognition and relationship extraction for questions, linking to the corresponding entities and relationships in the knowledge base, querying, matching and reasoning out candidate answers, and obtaining the target answer. Today, the reasoning process of intelligent question answering in the field of search engines mainly faces the following problems: [0003] 1) The traditional semantic analysis and information retrieval methods need to manually write a large number of templates or define a l...

AI Technical Summary

Problems solved by technology

[0003] 1) Traditional semantic analysis and information retrieval methods need to manually write a large number of templates or define a large number of rules

[0004] 2) Some deep learning methods can only deal with simple problems, cannot apply to complex reasoning processes, and consume a lot of computing resources

Images