| Abstract [eng] |
With a rapid development of technologies and the increasing number of internet users, more and more products and services are transferred into the virtual space. Various offers to buy something by means of internet or to use a certain service without leaving the home should save the client’s time. However, new problems of proper offers and choices occur here. Recommender systems are widely used to solve these problems. The aim of this work is to propose a new recommendation method, which determines specifics of user groups when user-item matrix has high density. The new method is suitable for datasets with large number of users and relatively small number of products. This type of datasets is very popular in specialized online stores and in the specific web directories. In the dissertation, an analytical review of the basic principles of the recommender systems had been performed; knowledge about the recommendation methods and their efficiency had been systematized; the experimental study of effectiveness of the popular recommendation methods had been performed; the new recommendation method that is suitable for high density datasets and determines specifics of user groups when generating recommendation was created. |
