Approach to hybrid recommender systems development

The article deals with the actual scientific and technical task of developing the architecture of a hybrid recommender system that uses domain knowledge, elements of a collaborative approach, as well as content data for a more complete and accurate analysis of the problem situation. Groups of criteria that implement each of these approaches were formulated and described. The main novelty in the approach proposed by the authors is the rejection of the initial focus on a specific subject area in favor of invariance. This approach provides a wider application of the development, reduces the cost of developing, debugging and implementing the system for the user.