Abstract
The article formulates a rigorously developed concept of opinion mining and sentiment analysis using hybrid neural networks. This conceptual method for processing natural-language text enables a variety of analyses of the subjective content of texts. It is a methodology based on hybrid neural networks for detecting subjective content and potential opinions, as well as a method which allows us to classify different opinion type and sentiment score classes. Moreover, a general processing scheme, using neural networks, for sentiment and opinion analysis has been presented. Furthermore, a methodology which allows us to determine sentiment regression has been devised. The paper proposes a method for classification of the text being examined based on the amount of positive, neutral or negative opinion it contains. The research presented here offers the possibility of motivating and inspiring further development of the methods that have been elaborated in this paper.
