Hybrid method for prediction of users’ information behavior in the Internet based on bioinspired search
Victoria V. Bova, Yu. A. Kravchenko, Sergey I. Rodzin, Кулиев Эльмар Валерьевич · Journal of Physics Conference Series · 2019
Abstract The paper presents the method of predicting the learner preferences to make recommendations for building the behavior profile for recommender systems (assistants) personalizing the educational activity in the Internet space. To form the personal information offer and to identify the groups of collaborative filtration of the users with similar preferences, the authors propose the hybrid model of collaborative filtration based on the Item-Item CF and User-User CF methods simultaneously. To reduce the size of space during the search process, the author present a novel heuristic bioinspired algorithm based on the fish school behavior in nature. This method can be characterized by the ability to scale and process the sparse data. To estimate the quality of the developed algorithm, experimental research was carried out on the basis of the benchmark dataset MovieLens.