Social media sentiment classification method and device based on knowledge graph

Abstract

The invention discloses a social media sentiment classification method and device based on a knowledge graph. The method comprises: firstly, based on construction of a basic knowledge graph, carryingout emotion entity positioning and emotion polarity labeling on entities in the basic knowledge graph in combination with an emotion dictionary, and constructing an emotion knowledge graph suitable for social media emotion classification tasks; then, using a GAN neural network model for improving the vector embedding precision of entities and entity relationships; training word vectors by using aCBOW model, and performing sentiment word vector training by using entity attributes in the knowledge graph; and finally, based on a Bi-LSTM multi-feature fusion sentiment classification strategy, fusing a general word vector, an entity vector and a sentiment word vector to the vocabulary input vector to improve the processing performance and precision of the sentiment classification method in thesocial media application. The method and device can effectively solve the common problems of low precision, complex operation and the like of a traditional sentiment classification method, and can bequickly and flexibly applied to social media sentiment classification.

Description

technical field [0001] The invention relates to a social media emotion classification method and device based on a knowledge graph, and belongs to the technical field of the Internet. Background technique [0002] With the continuous development and progress of science and technology such as electronic technology, computer technology, and Internet technology, the Internet has become the most important way for people to obtain information and resources. Among them, social media has also developed vigorously, and a large number of traditional media have settled in major social platforms, becoming an important source of high-quality content on social platforms. A large number of Internet users publish and disseminate tens of billions of information every day. A large part of these massive text information expresses the user's opinion tendency and emotional information. These emotional text information are very valuable opinion resources, which contain people's different views...

AI Technical Summary

Problems solved by technology

[0004] Purpose of the invention: Aiming at the problem that the current emotion classification technology requires manual labeling, the accuracy is not high, and it is difficult to cope with increasingly complex emotional expression methods, this invention proposes a social media emotion classification method and device based on knowledge graphs, which can build complete emotional knowledge Atlas, solve the common problems of low precision and complex calculation in traditional emotion classification methods, and improve the processing performance and accuracy of emotion classification methods in social media applications

Images