A Tibetan emotion analysis method and system based on mixed depth learning

Abstract

The invention relates to a Tibetan emotion analysis method and system based on mixed depth learning. The method comprises the following steps: S1, acquiring Tibetan sentences, and preprocessing the Tibetan sentences to obtain preprocessed Tibetan; S2, carrying out Tibetan word segmentation on that preprocessed Tibetan language to obtain the Tibetan language after word segmentation; S3, preliminarily processing the Tibetan language after word segmentation to obtain a word vector, and establishing a word vector model according to the word vector; S4, introducing that word vector model into a three-layer convolution neural network for iterative processing to extract the feature of the first word vector; S5, introducing that first word vector feature into a two-layer long-short-term memory network for iterative processing to extract the second word vector feature; S6, the second word vector feature is compiled and classified by a classification function. By using convolution layer and two-layer long-term and short-term memory network to analyze Tibetan language features, the global measure of text can be preserved and deeper semantic relations can be mined, and better classification results can be achieved.

Description

technical field [0001] The present invention relates to the field of computer application technology, in particular to a method and system for emotional analysis of Tibetan language based on hybrid deep learning. Background technique [0002] There are mainly three existing Tibetan sentiment analysis methods, one is based on the sentiment dictionary: the Tibetan sentiment analysis method based on the sentiment dictionary is to judge the Tibetan language by superimposing and calculating the weights of sentiment words or sentiment phrases in Tibetan sentences. emotional tendencies. Turning words in Tibetan sentences can change the emotional tendency of the whole sentence, so it is necessary to deal with turning words in Tibetan sentences. The degree words in Tibetan sentences can weaken or strengthen the strength of the emotional words or phrases in the sentence, so it is necessary to deal with the degree words in Tibetan sentences. Finally, the emotional tendency of Tibetan...

AI Technical Summary

Problems solved by technology

[0003] However, the above-mentioned Tibetan sentiment analysis method has the following defects: the Tibetan sentiment analysis method based on the emotional dictionary is the most basic method, and when calculating Tibetan sentences, it will lead to serious lack of semantics and low accuracy

The self-encoding based on deep learning also solves the Tibetan semantic problem to a certain extent, and the accuracy rate is not very ideal.

At present, domestic Tibetan sentiment analysis methods have solved the Tibetan semantic problem to a certain extent, but they are not very ideal in terms of sentiment analysis effect and semantic structure.

Images