The development of a web application for the automatic analysis of the tonality of texts based on machine learning methods

摘要

The work is devoted to the study of the existing methods of analyzing the sentimentality of the text, development and realization of web application for determining the key of the text in commentaries and paragraphs in social networks. Experiment is based on data sets from the social media portals like 'Tengrinews', 'Nur.kz' and 'Zakon.kz', which analyzes the tone of message entered by user. We developed a software to determine the tone of text data, according to the problem of the system for analyzing the emotional coloring of messages. For sampling data, we applied the data collecting system, called “Crawler”, which collects text data from our web portals. After assembling data from our portals, we apply several models of the algorithm for analyzing text data. In this study, models of the machine learning method are applied; the tone of the user input is analyzed, with the help of which we can get the social mood of the people. Thereby, the advantages of working with the above-mentioned methods are shown. A focus is to put on showing where and how naive Bayesian classifiers can provide valuable analysis in the text analysis process..