Deep-learning-based Chinese sentiment analysis method

Abstract

The invention discloses a deep-learning-based Chinese sentiment analysis method, comprising: acquiring online text, reasonably designing a conversion logic of Chinese sentences to mathematical vectors, constructing a word vector dictionary via a Chinese word segmentation technique in conjunction with a word vector learning tool, using an LSTM-MP (long short-term memory) model to perform sentence vector conversion, and using a softmax classifier to perform positive and negative sentiment classification on representative sentence vectors so as to arrive at sentiment analysis. The deep-learning-based Chinese sentiment analysis method has the advantages that classification accuracy is high, classification efficiency is high, flexibility is high, mass manual operations in supervised learning methods are avoided, text sentiment orientation classification efficiency and accuracy are effectively improved, automation integrity is high, and manpower sources are greatly saved.

Description

technical field [0001] The invention relates to a deep learning-based Chinese text sentiment analysis method, which belongs to the technical field of natural language processing and deep learning. Background technique [0002] The rapid development of the Internet has made Weibo and social networking popular forms of communication. Hundreds of millions of information reflecting people's opinions and attitudes are released and shared with everyone through platforms such as Twitter and Facebook every day, which provides opportunities to monitor and analyze the opinions and sentiments of private companies or social public domains. [0003] Text sentiment analysis is a type of technology that effectively analyzes people's opinions, emotions, attitudes, and emotional tendencies of entities such as products, services, organizations, and events, and then further conducts information inductive reasoning. Aiming at the massive data generated by the network media, extracting valuable...

AI Technical Summary

Problems solved by technology

However, due to the differences in the structure of English and Chinese texts, the research on sentiment analysis of Chinese texts is much more difficult. Coupled with factors such as late start, limited expectations for annotated texts, and backward technical methods, research on sentiment analysis of Chinese texts still has a lot to do. room for improvement

[0005] At present, most of the sentiment analysis methods for Chinese text are rule-based and supervised based on machine learning. The limitations include the following aspects: (1) Since the rules of language knowledge vary from person to person, the formulation of emotion judgment rules is limited by the research level of the makers; (2) Some methods artificially select features based on experience when extracting sentence features, so the effect of sentiment analysis is greatly affected by human factors, etc.

Images