ANN-based feature selection method in database text classification

Abstract

An ANN-based feature selection method in database text classification is characterized by comprising the following specific steps of using a text sample set and text categories of the text sample set,and obtaining a text classification model through ANN artificial neural network training; establishing a set of text classification models, and setting feature keywords for comparison for different text classification models; obtaining to-be-classified texts, preprocessing the to-be-classified texts, and obtaining a feature item set of the to-be-classified texts; determining the entity attributeof each feature item in the feature item set and the occurrence frequency of each feature item in the to-be-classified text; setting a weight according to the occurrence frequency of the feature item;sorting the feature items according to the degree of association; calculating the similarity; inputting the text to be classified into the text classification model with the highest similarity. According to the method, a plurality of classification models are obtained through ANN training, feature keywords of the classification models and feature items of texts to be classified are extracted, andthe most suitable classification model is selected through comparison.

Description

technical field [0001] The invention relates to the technical field of computer text classification, in particular to a feature selection method in ANN-based database text classification. Background technique [0002] ANN refers to a complex network structure formed by the interconnection of a large number of processing units (neurons), which is a certain abstraction, simplification and simulation of the organizational structure and operating mechanism of the human brain. Artificial Neural Network (ANN for short), which simulates neuron activity with a mathematical model, is an information processing system based on imitating the structure and function of the brain's neural network; artificial neural networks can be divided into multi-layer and single-layer, each One layer contains several neurons, and each neuron is connected by a directed arc with variable weight. The network can process information, simulate The purpose of the relationship between input and output. It d...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the database field existing in the background technology, it is essential to classify the database text, the classification method in the prior art is relatively cumbersome, and the technical problem of extracting the characteristic words in the text is relatively troublesome, the present invention proposes a method based on The feature selection method in the database text classification of ANN, obtains a plurality of classification models by ANN training in the present invention, and extracts the feature keyword of classification model and the feature item of text to be classified, selects the most suitable classification model by comparison, Classification results are more accurate

Images