Design and Development of the Recommendation System for a Dating Search Application

This thesis is devoted to the design and development of a recommendation system in the Yume dating search application, focused on the formation of long-term relationships based on the psychological compatibility of users. The introduction to the work contains an analysis of the current situation in the online dating market, including problems and opportunities related to the development of technology and changes in user needs. The main problem highlighted in the work is the insufficient effectiveness of existing solutions for creating long-term relationships, despite the growing demand for such services. The aim of the work is to develop an architecture of a recommendation system that takes into account psychological factors in order to increase the likelihood of establishing meaningful connections in real life for users of the Yume application. To achieve this goal, the following main tasks were set: Formation of requirements for the recommendation system. Analysis of architecture components and methods of their development. Implementation of the system and preparation for implementation. Justification of the effectiveness of the developed system. The main results of the work include: Development of a recommendation system architecture that takes into account the psychological parameters of users. Evaluating the effectiveness of the system based on qualitative metrics. Designing and implementing the backend of the system based on the Python language and the FastAPI framework. Selecting and configuring a MongoDB database to store user interaction data. Conducting alpha and beta testing of the system, as well as training a gradient boosting model to improve the quality of recommendations. The work represents a significant contribution to the development of recommendation systems for online dating, offering an innovative approach to taking into account the psychological compatibility of users.