Natural language model generation method and computer equipment

Abstract

The invention relates to a natural language model generation method and computer equipment, and the method comprises the steps: inputting a shielding statement pair group into an initial neural network, so as to obtain a prediction statement pair group and a prediction label, wherein the shielding statement pair group is obtained by preprocessing statement pairs in training data, and the trainingdata comprises the statement pairs and real labels; and adjusting parameters of the initial neural network according to the statement pair, the real label, the prediction statement pair group and theprediction label, and continuing to execute the step of inputting the shielding statement pair group into the initial neural network until a preset training condition is met so as to obtain a trainednatural language model. According to the natural language model obtained through adoption of the method, semantic representation of the natural language model has global information and local semanticinformation of the shielded statement pair, so that the precision of a natural language processing task is improved.

Description

technical field [0001] The present application relates to the technical field of image processing, in particular to a method and computer equipment for generating a natural language model. Background technique [0002] Natural Language Processing (Nature Language Processing, NPL) is a subfield of artificial intelligence, usually divided into four categories of tasks: sequence labeling, classification tasks, relationship judgment, and generative tasks, training natural language processing models to improve natural language processing The accuracy of the task results of the NLP plays an important role, because the appropriate word vectors are obtained through training, and the appropriate word vectors can improve the accuracy of the task results of natural language processing. [0003] NLP generally uses certain tasks in natural language processing for training. For example, the neural network language model (Nature Network Language Model) proposed by Bengio, the mainstream me...

AI Technical Summary

Problems solved by technology

[0003] NLP generally uses certain tasks in natural language processing for training. For example, the neural network language model (Nature Network Language Model) proposed by Bengio, the mainstream methods of CBOW and Skip-gram training proposed by Google, mostly by inputting a certain The context-related words of a feature word can learn the meaning information of the feature word. These mainstream training methods do not learn the word meaning information and global semantic information of the associated words adjacent to the feature word. Therefore, through the existing training methods, we can get The language processing model of the isolated word cannot learn the associated word meaning information before and after the isolated word, as well as the global semantic information. The semantic representation ability is weak, resulting in low precision of the processing results.

Images