Semantic comprehension system and method oriented to Chinese text

Abstract

The invention provides a semantic comprehension system and method oriented to a Chinese text. Based on deep learning, a deep learning text classification model is provided; the model is divided into an input layer, a convolutional layer, a pooling layer, a GRU (Gated Recurrent Unit) layer, a fully connected layer and an output layer; a pinyin characteristic sequence of text segmentation is used asinput; characteristics are obtained through multi-layer characteristic extraction; and an intention category is predicted to obtain a text classification result. According to the semantic comprehension system and method oriented to the Chinese text, the part of speech of a statement does not need to be judged, a complex preprocessing process such as a syntax analysis tree and the like does not need to be generated, the text only needs to be segmented and the segmented text is converted into pinyin, and the problems that the relation between words and words cannot be measured, a lot of external prior knowledge is needed and the curse of dimensionality is easily generated when large-scale corpuses are processed in a conventional characteristic extraction method are solved.

Description

technical field [0001] The present invention relates to the field of natural language processing. Specifically, it can be widely applied to any intelligent terminal that requires man-machine voice interaction and semantic understanding, such as unmanned vehicles, smart homes, and robots. Background technique [0002] With the development of social economy and science and technology, human beings have entered the era of artificial intelligence. A large number of intelligent products such as unmanned vehicles, smart phones, smart TVs, and service robots have entered people's lives and changed human life. In order to make smart products better serve humans, users need to interact with them in real time, and smart terminals need to understand the user's intentions. Among the many interaction methods, voice-based interaction is the most convenient, most convenient, and most widely used interaction method. Voice interaction has become the most important interaction method between ...

AI Technical Summary

Problems solved by technology

A method based on statistics such as the invention patent (Notice No.: CN104408095A) is an improved KNN text classification method, which first generates a vector space model of the training set, defines the sample space as a plurality of spherical areas by type, and then tests according to the distance To judge the category, this method is suitable for text space classification with high feature dimension, but when the number of categories increases, the classification accuracy of the space classification surface will decrease; the invention patent (notification number: CN105912716A) is a text classification method based on SVM , which introduces the concept of extended words, which overcomes the problem of sparse features of short texts, but it will cause the problem of excessive dimensionality in the case of large-scale corpora; methods based on deep learning such as invention patents (notification number CN104834747A) are a kind of The classification method based on word vector and convolutional neural network maps words to a point in space and uses the distance between them to measure the relationship between words, and uses convolutional neural network to extract features, but it is difficult to deal with complex intent text or When the statement is long, the context information will be lost

Images