Detecting Negative Content on the Internet Using Machine Learning

Object of development is an algorithm for detecting negative text content and a program module that implements the developed algorithm. The purpose of the work is review of machine learning methods used in the task of detecting negative text content, and development of software module to address this task. In the course of the work, a review of literary sources on the topic of text classification was made. The algorithm for detecting negative text content with possibility of work with texts of any length and with use of text preprocessing methods, feature extraction and text classification methods was developed. Program module was created on its basis. Computational experiments were made and based on them the best methods were compared and revealed, the algorithm was compared with existing analogues, the possibility of work with texts of any length was studied. The results can be applied in the task of automated analysis of web-content. Explanatory note contains 45 pages, 7 pictures, 10 tables, 27 sources, 1 appli-cation.