A user emotion analysis method based on deep learning

Abstract

The invention discloses a user emotion analysis method based on deep learning. The method comprises following steps of Constructing a vocabulary; Using a classifier to classify and select one part asa seed dictionary, giving emotion polarity score dictionaries of the seed corpora, and improving the corresponding TF-IDF feature selection process; Storing the result of the improved TF-IDF feature word vector and the result of the Word2vec feature word vector; Summing and averaging the emotion seed words of each emotion classification to obtain the probability of the polarity of the central wordvector of each emotion. The invention provides a user emotion analysis method based on deep learning. and word frequency information, emotion information and semantic information of the text are effectively combined by combining a basic word frequency feature selection algorithm and a Word2vec algorithm, and a comparison test is carried out on a newly constructed emotion semantic word vector andan original word vector, so that the effectiveness of the emotion semantic word vector is effectively proved.

Description

technical field [0001] The invention relates to the technical field of sentiment analysis, in particular to a user sentiment analysis method based on deep learning. Background technique [0002] As a platform for people's information exchange and resource sharing, the Internet saves a large amount of data containing subjective information. How to extract people's interests, texts with opinions, and sentiment classification from these massive data is one of the current research hotspots. one; [0003] Traditional methods for solving sentiment analysis problems include unsupervised methods based on sentiment dictionaries and human judgment rules, and supervised methods based on machine learning. These methods can achieve certain results when the amount of data is not large or the semantics are not rich enough. , but as the amount of data becomes larger and the expression methods become more and more abundant, the traditional methods can no longer effectively solve this kind o...

AI Technical Summary

Problems solved by technology

[0003]Traditional methods for solving sentiment analysis problems include unsupervised methods based on sentiment dictionaries and artificial judgment rules, supervised methods based on machine learning, and These methods can achieve certain results when the data size is large or the semantics are not rich enough, but as the amount of data becomes larger and the expression methods become more and more abundant, traditional methods can no longer effectively solve this type of problem, and new methods are urgently needed. propose