Solving the problem of sentiment analysis using neural network models

Abstract

© 2020, Advanced Scientific Research. All rights reserved. The article considers methods that create a vector representation of words in the n-dimensional vector space in order to solving the problem of sentiment analysis based on neural network models of natural language processing. The methods are based on "Word2Vec", "GloVe", "FastText" technology. Approaches are used in the tasks of classification, sentiment analysis, typo correction, recommendation systems. We present the results of classifications comparison in the problem of sentiment analysis of a multilayer perceptron, a convolutional and recurrent neural network, decision trees (random forest), support vector machine (SVM), naive Bayes classifier (NB), and k-nearest neighbors (K-NN). The results of the classification are presented for three data sets: Twitter messages, reviews of various goods and services, Russian-language news.