Emotion analysis method and device based on big data and deep learning and storage medium

Abstract

The invention discloses an emotion analysis method and device based on big data and deep learning and a storage medium. The emotion analysis method comprises the steps of preprocessing emotion data with preliminary labeling information; adopting a heuristic extending mode to label the emotion data, wherein the heuristic extending mode comprises the procedures of conducting word-splitting on the emotion data, extracting emotion words, and distributing emotion strength values to the emotion words; extracting adjunct words and distributing adjunct strength values to the adjunct words; calculatingthe emotion strength of phrases; adding the phrases and the emotion strength values to an emotion labeling dataset if the emotion polarity corresponding to the emotion strength of the phrases is consistent with that of the emotion data. By means of the emotion analysis method and device based on the big data and deep learning and the storage medium, and by heuristically extending the collected emotion data, good original data can be provided for subsequent data processing, the pertinency and accuracy of subsequent data analysis are improved, so that an emotion lexicon is more accurate and complete.

Description

technical field [0001] The present invention relates to the field of computer sentiment analysis, in particular to a sentiment analysis method, sentiment analysis device, storage medium, computer equipment and program products. Background technique [0002] With the rapid development of the Internet, especially the mobile Internet, various new applications are constantly emerging, such as news portals, e-commerce websites, social networks, and so on. On these applications, the general public can express their subjective views on things, such as comments on news events and preferences for purchased products. Currently, user-generated text data with sentimental tendencies is growing exponentially. By mining these massive data, obtaining user emotional information is crucial to e-commerce, business intelligence, public opinion surveys, public opinion analysis, intelligence analysis, enterprise management, etc., and also provides powerful decision-making support for managers. ...

AI Technical Summary

Problems solved by technology

However, under normal circumstances, dynamic emotional words will have a great limiting effect on emotional words such as adjectives, and even affect the part of speech of emotional words. If they are not considered, it is easy to cause deviations in emotional intensity, thereby affecting the accuracy of data labeling in emotional data sets. which in turn affects the results of text reviews

Therefore, the accuracy of existing sentiment analysis models is often relatively low

Images