No abstract

No abstract

Sentiment analysis is a key task in natural language processing and has a wide range of real-world applications. Traditional methods classify "plain texts" as "positive" and "negative". We propose a method that is significantly different from traditional approaches. In addition to "plain text", we have analyzed ways to express emotions in a message and a variety of emotional indicators, emoticons and emojis. The proposed model with emotional indicators to predict text polarity improves the prediction accuracy of sentiment classes by 6% compared to traditional models. The model uses an original data set marked up according to an extended list of Plutchik emotion classes. Therefore, the model predicts 8 independent sentiment classes: "joy", "sad", "distaste", "fear", "anger", "surprise", "attention" and "trust". The novelty of the research also lies in using the Russian language data set. The model considers national linguistic, semantic and semiotic features of social environment.

No abstract

No abstract