1: \begin{abstract}

2: Recommender systems have been widely adopted by electronic commerce and entertainment industries 

3: for individualized prediction and recommendation, which benefit consumers and improve business intelligence. 

4: In this article, we propose an innovative method, {\color{black} namely the recommendation engine of multilayers (REM),} for tensor recommender systems.

5: The proposed method utilizes the structure of a tensor response to integrate information from multiple modes,

6: and creates an additional layer of nested latent factors to accommodate between-subjects dependency. One major advantage is that the proposed method is able to address the ``cold-start'' issue in

7: the absence of information from new customers, new products or new contexts. Specifically,

8: it provides more effective recommendations through sub-group information. 

9: %In theory, the proposed method achieves an optimal convergence rate under the $L_2$-loss function and a fast convergence rate in more general settings. 

10: To achieve scalable 

11: computation, we develop a new algorithm for the proposed method, which incorporates a

12: maximum block improvement strategy into the cyclic blockwise-coordinate-descent algorithm. 

13: {\color{black} In theory, we investigate both algorithmic properties for global and local convergence, along with the asymptotic consistency of estimated parameters.}

14: Finally,

15: the proposed method is applied in simulations and IRI marketing data with 116 

16: million observations of product sales. Numerical studies demonstrate that the proposed method outperforms existing competitors in the literature.

17: %TO DO: Add new things about theory

18: \end{abstract}

19: