Statement classification model training method and device

Abstract

The invention relates to a statement classification model training method and device, a computer readable storage medium and computer equipment, and the method comprises the steps: obtaining a searchstatement set which comprises a plurality of to-be-replied search statements; performing screening from the search statement set according to a preset screening rule to obtain a candidate search statement set; inputting the candidate search statement set into the original statement classification model to obtain domain search statements corresponding to the concerned domain; searching a preset search statement library to obtain a target search statement matched with the domain search statement; taking the candidate answer corresponding to the target search statement as a domain answer corresponding to the domain search statement to obtain a domain question and answer pair corresponding to the concerned domain; and inputting the domain search statement corresponding to the domain question and answer pair into the original statement classification model for training until a convergence condition is met to obtain a target statement classification model. According to the scheme provided bythe invention, the search statement reply accuracy can be improved.

Description

technical field [0001] The present application relates to the field of computer technology, in particular to a sentence classification model training method, device, computer-readable storage medium and computer equipment. Background technique [0002] With the development of computer technology, man-machine question answering technology has appeared, which replaces or replaces human beings to answer user inquiries by man-machine question answering technology. However, the current question-and-answer pairs of training data for the human-machine question-answering model are all question-answer pairs that need to be crawled from a large number of web pages, and then used as the training data of the human-machine question-answering model for training to obtain a trained human-computer question-answering model. Since the question-answer pairs crawled from the webpage have serious quality problems, it is easy to cause the human-machine question-answering model to have a low accur...

AI Technical Summary

Problems solved by technology

However, the current question-and-answer pairs of training data for the human-computer question answering model are all question-answer pairs that need to be crawled from a large number of web pages, and then used as the training data of the human-computer question answering model for training to obtain a trained human-computer question answering model. Due to the large quality problems of the question-answer pairs crawled from the webpage, it is easy to cause the human-computer question-answer model to have a low accuracy rate when replying to search sentences.

Images