Analysis of the preferences of public transport passengers in the task of building a personalized recommender system

Abstract

The paper presents the theoretical and algorithmic aspects for making a personalized recommender system (mobile service) designed for public route transport users. The main focus is on identifying and formalizing the concept of "user preferences", which is the basis of modern personalized recommender systems. Informal (verbal) and formal (mathematical) formulations of the corresponding problems of determining "user preferences" in a specific spatial-temporal context are presented: the preferred stops definition and the preferred "transport correspondence" definition. The first task can be represented as a well-known classification problem. Thus, it can be formulated and solved using well-known pattern recognition and machine learning methods. The second is reduced to the construction of dynamic graphs series. The experiments were conducted on data from the mobile application "Pribyvalka-63". The application is the tosamara.ru service part, currently used to inform Samara residents about the public transport movement.