We analyse the state-of-the-art recommendation
systems based on the tensor factorization techniques.
The comparison of these systems is provided,
the possible problems and the drawbacks
are identified. We developed the HOSVD recommendation
system. Using the implemented recommender
system the experiments are conducted
and the results are reported. The majority of the
identified problems are verified during the experiments.
Moreover, we propose a number of the
possible solutions and improvements to the problems.
Finally, we describe the main goals of our
work during the next semester.