Intelligent question-answer method and system

Abstract

The invention provides an intelligent question-answer method and system. The method comprises the steps that questions and corresponding answers stored in a knowledge base are obtained, and agentive and passive persons of the questions are calculated according to a person model; the similarity between keywords of questions of a user and keywords of the questions in the knowledge base is calculated, a first question set is obtained, and the agentive and passive persons corresponding the questions in the first question set are obtained in the knowledge base; agentive and passive persons corresponding to the question of the user are calculated; the questions with the agentive and passive persons inconsistent with the agentive and passive persons corresponding to the question of the user are removed, and a second question set is obtained; one question in the second question set is randomly obtained to serve as a matching question, and the answer corresponding to the matching question serves as the answer of the question of the user. The agentive and passive persons in the question raised by the user are calculated through the person model, the questions with the agentive and passive persons inconsistent with the agentive and passive persons of the question raised by the user are removed, the similar question is matched, and the accurate answer is given.

Description

technical field [0001] The invention relates to the technical fields of electrical digital data processing and artificial intelligence, in particular to an intelligent question answering method and system. Background technique [0002] In intelligent dialogue systems, the method of matching questions and questions is usually used to find answers. Save questions and corresponding answers in the knowledge base. When a user asks a question A, find a question B similar to question A from the knowledge base, and then return the answer to question B to the user. Usually, the similarity between two questions is calculated by keyword comparison, that is, the similarity is calculated based on the keywords of question A and question B. In order to improve recall, an exact keyword match is usually not required, however, this approach may introduce errors. Because the keywords extracted in the two questions may have a high degree of similarity, but the order of the persons contained i...

AI Technical Summary

Problems solved by technology

In order to improve the recall rate, the keyword is usually not required to match exactly, however, this method may introduce errors

Because the keywords extracted in the two questions may have a high degree of similarity, but the order of the persons contained in the two questions is different in the sentence, that is, the agent person and the subject person are different, and the agent person in question A is called X1, the subject is called Y1, and the agent in question B is called X2, and the subject is called Y2, so if you ignore the problem that the agent and subject are different in the question, it is possible to give the answer to the question is not accurate

For example, for example, question A: I like you question B: You like me, the keywords of these two questions are "I" "like" "you", so there will be a high degree of similarity, but because the two sentences The agent person and the affected person in are different, resulting in completely different semantics, so it is likely to give an inaccurate answer to the question

[0003] Therefore, the defect in the prior art is that in the intelligent dialogue system, it is impossible to perform semantic analysis according to the actors and recipients included in the question, resulting in inaccurate answers

Images