Research into Development Principles of a Recommender System Based on Navigational and Behavioral Patterns of Users on a Real Estate Site

Modern recommender algorithms are based on approaches and algorithms designed to work in the e-commerce and online entertainment domains. Most of these algorithms use explicit user actions to personalize recommendations. In most cases, such systems are aimed at creating recommendations for registered users with a long history of activity. However, there are domains where the application of such approaches is complicated. In this context, the present thesis is devoted to studying the development principles of recommender systems for the real estate domain. The standard recommender algorithms are explored; their shortcomings and limitations are evaluated within the scope of the present research. Based on the considered algorithms, a hybrid solution taking into account the features of each of the internal algorithms is proposed. The master's paper consists of 60 pages, 3 chapters, list of references (41 sources), introduction, and conclusion. Keywords: hybrid recommender system, real-estate, collaborative-filtering, content-based filtering, business rules filtering, item similarity, implicit recommendations.