Text matching method and device based on prediction model, equipment and storage medium

Abstract

The invention relates to the technical field of artificial intelligence, and particularly discloses a text matching method and device based on a prediction model, equipment and a storage medium. The method comprises the steps of obtaining a pre-training sample; inputting the pre-training sample into a pre-constructed prediction model to obtain a semantic vector of each word, and performing mask mark prediction and text sample similarity prediction according to the semantic vectors to obtain a first prediction result and a second prediction result; calculating a target loss function according to the first prediction result and the second prediction result, and training a prediction model by using the target loss function to obtain a target prediction model; and obtaining a to-be-predicted text containing two to-be-matched texts, inputting the to-be-predicted text into the target prediction model, obtaining a similarity prediction result of the two to-be-matched texts, and determining whether the two to-be-matched texts are matched or not according to the similarity prediction result. In this way, the generalization ability of the model can be improved, the over-fitting risk is reduced, and the training efficiency is improved.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to a predictive model-based text matching method, device, equipment and storage medium. Background technique [0002] With the increasing application of intelligent customer service in telephone scenarios in recent years, the intersection between people and intelligent customer service in daily life is increasing. [0003] In intelligent dialogue, the development of the model for identifying customer intent has gone through several rounds of iterations, from the initial tf-idf, to word vectors, to LSTM, and recently to the pre-training model represented by bert. When using pre-trained models such as bert, the general practice is to fine-tune them directly on the current task data set. However, the shortcoming of this solution is that since the amount of data in the task data set is generally relatively small, after several rounds of training, serious overfit...

AI Technical Summary

Problems solved by technology

[0004] The present invention provides a text matching method, device, equipment and storage medium based on a predictive model, which can solve the problem of serious overfitting phenomenon and low training efficiency caused by traditional model round-robin training for a single task

Images