Natural language theme classification method and device

Abstract

The invention discloses a natural language theme classification method and device. A training stage in the method comprises the following steps: acquiring a natural language text segment with a knowntheme as a sample set; extracting a plurality of words with the highest occurrence frequency in the sample set to obtain a plurality of feature words; representing each feature word as a vector to obtain a plurality of feature vectors; calculating the similarity degree between any two feature vectors to obtain a similarity degree set; inputting the similarity degree, the topics and the feature words corresponding to the topics into a preset neural network structure for training to obtain a feature spectrum and a model for expressing a relationship between the feature spectrum and a classification result. The classification stage comprises the steps of obtaining a to-be-classified natural language text segment; extracting feature words belonging to the feature spectrum in the to-be-classified natural language text segments to obtain input feature parameters; and inputting the input characteristic parameters into a model for expressing the relationship between the characteristic spectrumand the classification result to obtain the classification result. The method can achieve the adaptive selection of features, and improves the classification accuracy.

Description

technical field [0001] The invention relates to the field of natural language classification, in particular to a natural language subject classification method and device. Background technique [0002] Text classification algorithms have existed for a long time. In the 1950s, scientists had used the "expert system" method to classify texts. However, the scope of coverage and classification accuracy of this method are very limited, and it can only be used to solve some problems. Well-conditioned, well-described, and organized text classification problems. With the development of statistical methods, especially the increase in the number of online texts on the Internet and the rise of machine learning disciplines after the 1990s, a set of classic methods for solving large-scale text classification problems has gradually formed. The main process is "artificial feature engineering" + "Classification model", which splits the entire text classification problem into two parts: fea...

AI Technical Summary

Problems solved by technology

However, the establishment of the initial feature library of this method relies on manually selected features, which cannot achieve adaptive selection of features, and the classification accuracy is limited.

Images