Analysis of User Feedback from the Application to Develop Recommendations for Developers Using Machine Learning Methods

With the increasing use of smartphones in everyday life, desire of various businesses to develop in this area and write new applications is also growing. As competition increases, developers are forced to use the iterative process of developing, testing, and improving the quality of their applications. Therefore, good and timely feedback from users becomes very valuable for the business, in particular for people who are involved in the further development of the application and its support. Using the information obtained from the feedback, developers can successfully identify and fix bugs, implement new functionality, which is ultimately determined directly by people using the application, improve the quality of user expectations and experience interacting with the application. This paper discusses the relevance of the analysis of reviews and their characteristics. Methods of analysis of text information are also considered, techniques for text preprocessing, vectorization of textual information are analyzed, and a comparative analysis of classifiers is carried out. In addition, existing methods for grouping text documents by topic will be described. Also, recommendations were created for developers in the form of “informational” reviews grouped by topic, sorted by the importance that the developed ranking model will determine. The final part describes in detail the algorithm of the ranking model and the results obtained.