Proceedings of the 2nd International Scientific and Practical Conference “Modern Management Trends and the Digital Economy: from Regional Development to Global Economic Growth” (MTDE 2020)

Designing a Digital Content Recommendation System for films

Authors
A.V. Zagranovskaia, D.Yu. Mitiura, T.A. Makarchuk
Corresponding Author
A.V. Zagranovskaia
Available Online 5 May 2020.
DOI
https://doi.org/10.2991/aebmr.k.200502.001How to use a DOI?
Keywords
recommendation systems, content filtering, data analysis, vectorization of descriptions method, latent semantic analysis method, proximity metrics, factor analysis
Abstract
Purpose of the study. The purpose of the study is to analyze the existing methods for constructing content recommendation systems and developing the most accurate and adequate content recommendation system for films. Materials and methods. The paper considers a variety of data analysis methods: vectorization of descriptions, latent semantic analysis (LSA) and its probabilistic form (pLSA), latent Dirichlet allocation (LDA), proximity metrics (cosine divergence, Jensen-Shannon divergence, Kullback-Leibler divergence), the most common factorization method is truncated SVD. Designed content recommendation systems are tested on MovieLens dataset open data [1]. This opens up opportunities for checking the results obtained and improving the proposed models. Results. As a result, several models of content recommendation systems for films were developed and reviews of potential users of the system were analyzed which allowed determining the best version of the content recommendation system. Conclusion. An analysis of the created recommendation systems made it possible to understand the user requirements for them. The most important criterion was the level of trust to the system. In other words, this is how much the user is sure that he or she will really like offered recommendations. The quality of the recommendations was evaluated using a survey of the system potential users. This made it possible to cover a number of criteria for evaluating recommendation systems, such as accuracy of rating prediction, novelty, surprise and diversity.
Open Access
This is an open access article distributed under the CC BY-NC license.

Download article (PDF)

Cite this article

TY  - CONF
AU  - A.V. Zagranovskaia
AU  - D.Yu. Mitiura
AU  - T.A. Makarchuk
PY  - 2020
DA  - 2020/05/05
TI  - Designing a Digital Content Recommendation System for films
BT  - 2nd International Scientific and Practical Conference “Modern Management Trends and the Digital Economy: from Regional Development to Global Economic Growth” (MTDE 2020)
PB  - Atlantis Press
SP  - 1
EP  - 8
SN  - 2352-5428
UR  - https://doi.org/10.2991/aebmr.k.200502.001
DO  - https://doi.org/10.2991/aebmr.k.200502.001
ID  - Zagranovskaia2020
ER  -