Network context text recognition method and device and storage medium

Abstract

The invention provides a network context text recognition method and device and a storage medium. The method comprises the steps: constructing a style semantic model based on a text long window, and constructing a component semantic model based on a text short window; performing training based on the style semantic model vector model and the component semantic model by using a corpus of the network context to obtain a Chinese word vector model of the network context; and using the Chinese word vector model of the network context to identify an input text of the network context and outputting an identification result. According to the method, two different windows are used during word segmentation; the long window is used for extracting semantic information of a networked style; the text short window is used for extracting semantic features with different fine grit, the long window and the short window are combined in the training stage,more accurate word vector expression is obtained,so that the text recognition rate of the network context is improved. According to the method, a target function is optimized, the model training speed is increased, a radical meaning transfer methodis established during training, and therefore the text recognition rate is increased.

Description

technical field [0001] The invention relates to the technical field of text data processing, in particular to a text recognition method, device and storage medium in a network context. Background technique [0002] Text vector representation has always been an important direction in the research of computer technology and artificial intelligence technology, and it is also a major challenge in natural language analysis and processing. The quality of text vectorization representation directly affects the performance of subsequent natural language analysis models. The vectorized representation of text first adopted the one-hot encoding model, which gradually evolved into the Bag of Words model. These representation methods are simple and clear in thinking, and they better solve the problem of representing text in computers. problem, but does not consider the semantic correlation between word contexts in language and the time series characteristics of language, which splits the...

AI Technical Summary

Problems solved by technology

However, compared with English, the semantic composition of Chinese is more complicated. Chinese words are composed of Chinese characters, and the semantics of Chinese characters are generally related to the meaning of their word-forming components. If the CBOW model is directly used to learn Chinese word vectors, the potential semantics of Chinese characters will be ignored. information, the generalization ability of the word vector model is weak

And the recognition of text is not necessarily effective, the convergence speed of the model is slow during training, etc., there is an urgent need for a new Chinese word vector model to solve one or more of the above technical defects

Images