A News Classification Method Based on Text Context Structure and Attribute Information Overlay Network

Abstract

The invention discloses a newspaper news classification method based on text context structure and attribute information superposition network, belonging to the field of information processing. The present invention uses the text vector representation method to convert the text of indefinite length into a fixed-length vector to avoid the loss and redundancy of text information; from the perspective of training data, weighted random sampling is adopted to adjust the possibility of sample selection through weight , to optimize the composition of training samples; from the perspective of feature extraction, this paper not only considers the contextual result information but also text attribute information to optimize the feature extraction process. The present invention not only improves the extraction method of text features, but also additionally incorporates attribute features into the process of feature construction. Using the text vector representation method to convert the variable-length text into a fixed-length vector avoids the loss and redundancy of text information, and optimizes the extraction method of text features; additional feature information of news is added to enrich the source of features.

Description

technical field [0001] The invention belongs to the field of information processing, and relates to a news classification method and system based on a superimposed network of text context structure information and attribute information. Background technique [0002] Key term definitions: [0003] Neural network: It is a mathematical model or computational model that imitates the structure and function of biological neural networks, and is used to estimate or approximate functions. Neural networks are computed by a large number of artificial neuron connections. In most cases, the artificial neural network can change the internal structure on the basis of external information, which is an adaptive system. [0004] Text representation: It is a machine learning technique in the field of natural language processing that maps text, a high-level cognitive abstract entity, to a vector on the real number field for subsequent computer processing. [0005] Weighted random sampling: ...

AI Technical Summary

Problems solved by technology

Finally, for newspaper and periodical news, although the main reason for whether it can become the front page news is the amount of information contained in the context of the news text, but due to the limitations of the front page layout and later typesetting, some news that is too long or too short hard to make front page news

Most of the previous technical means only considered the text context structure information, while ignoring the text attribute information such as "title length" and "text length", resulting in the lack of features

Images