Machine intelligent question and answer method and device and electronic equipment

Abstract

The embodiment of the invention provides a machine intelligent question and answer method and device and electronic equipment. The method comprises the steps that if it is detected that a received message is a questionnaire, a text recognition model with a weight is adopted, intention recognition and field recognition are carried out based on the received message, so that the questionnaire of a target field is screened, and specific questions of the questionnaire of the target field are recognized; based on the specific question, an answer corresponding to the specific question are obtained and output by inquiring an intelligent question and answer knowledge base of the target field, wherein the text recognition model with the weight is used for endowing different segmented words in the received message with different weight coefficients. According to the embodiment of the invention, the question recognition and the text recognition model with the weight and attention mechanism are combined, and only the questionnaires in the target domain are screened for intelligent answering, so that the effectiveness, accuracy and efficiency of intelligent question answering can be effectivelyimproved when intelligent answering is performed based on the knowledge base in the target domain.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of artificial intelligence, and more specifically, relate to a machine intelligent question answering method, device and electronic equipment. Background technique [0002] With the rapid development of NLP technology, various chatbots are emerging, and they have achieved good results in small chat and task-based dialogue. The current machine question answering system generally adopts the question answering mode based on the knowledge base. When performing answer retrieval based on the knowledge base, an important premise is to accurately identify the questions entered by the user, so as to determine the user's true intentions and output answers that match the user's questions more closely. [0003] When identifying user questions in the prior art, word segmentation processing is usually performed on user question segments first, and then user questions are identified by separately ide...

AI Technical Summary

Problems solved by technology

However, in the process of using this recognition method for problem recognition, sometimes there will be the following problems: for a segment that expresses the same meaning, there will be a large deviation in the recognition result due to the difference in several words or even punctuation marks. This will seriously affect the validity and accuracy of the output answer

Images