A fuzzy cold-start recommender system for educational trajectory choice

Several approaches to choosing an educational trajectory are considered, and the advantages of using recommender systems are determined. The cold start problem of recommender systems is formulated and solved by creating a hybrid recommender system that combines a rule-based fuzzy expert system and a recommender system with fuzzy collaborative filtering. As one application, the general approach is implemented for choosing the field of study when entering a higher education institution. A modification of Klimov's career guidance test is used as initial data. The rules for estimating the metrics and similarity of fuzzy triangular data are presented. The algorithms of a fuzzy expert system and a fuzzy recommender system with collaborative filtering are described in terms of the fuzzy representation accepted. The two approaches are combined by generating pseudo data using an expert system. This provides a solution of the cold start problem and yields a recommender system whose quality is gradually improved by substituting the values from real user queries into the database. The programs implementing these algorithms are tested to confirm the effectiveness of the fuzzy recommender system.