Hybrid filtering and semantic sentiment analysis by deep learning for recommendation systems

  • Badiaa Dellal-Hedjazi Houari Boumediene USTHB University
  • Zaia Alimazighi Houari Boumediene USTHB University


Faced with the ever-increasing complexity, volume and dynamism of online information, recommendation systems are among the solutions that anticipate the needs of users and offer them items (articles, products, web pages, etc.) that they are likely to appreciate. Unlike traditional recommendation models, deep learning offers a better understanding of user requests, characteristics of objects, historical interactions between them and it can process massive amounts of data. In this work we realize a recommendation system based on MLP deep learning adapted to data already defined by their characteristics. In addition to the use of deep learning, we offer a new hybrid recommendation system solution between the demographic approach and the content-based approach in order to eliminate the limits of each and to combine their strengths, through a deep neural network that harnesses the mass of data. "Experimentation with our approach has produced good results in terms of accuracy and speed, whether through the use of deep learning or the hybridization of content-based and demographic filtering, which is a particular case of collaborative filtering.
Sep 16, 2022
How to Cite
DELLAL-HEDJAZI, Badiaa; ALIMAZIGHI, Zaia. Hybrid filtering and semantic sentiment analysis by deep learning for recommendation systems. International Journal of Information Science and Technology, [S.l.], v. 6, n. 3, p. 16 - 28, sep. 2022. ISSN 2550-5114. Available at: <https://innove.org/ijist/index.php/ijist/article/view/182>. Date accessed: 27 nov. 2022. doi: http://dx.doi.org/10.57675/IMIST.PRSM/ijist-v6i3.182.
Special Issue : Machine Learning and Natural Language Processing