Abstract
The recommendation technologies are used as viable solutions for advertising the best products and for helping users to orientate themselves in large e-commerce platforms offering various product assortments. Despite their popularity they still suffer of cold start and sparse data matrices limitations, which affect seriously the effectiveness of recommenders employed in applications with less user-system interaction. Having the aim to improve the quality of recommendation lists in such systems we introduce time heuristics into the recommendation process and propose two new variants of collaborative filtering algorithms for solving these problems. A time aware method is proposed for making more correct evaluations of recommenders used in domains with strong time dependencies.