Deep learning-based multi-context-aware multi-criteria recommender system

As digital information becomes ubiquitous, Recommender Systems (RSs) become increasingly critical in navigating this vast landscape. In this dissertation, we propose a novel deep learning-based approach to integrate multi-context and multi-criteria data into a unified neural network model. A context-aware and multi-criteria recommender system enhances traditional two-dimensional methods of RSs by incorporating context awareness and multiple criteria. In our work, data from multiple contexts and multiple criteria are intricately woven together within a single architecture, unlike traditional approaches. Concurrent processing allows sophisticated interactions between context and criteria, resulting in more accurate recommendations. When context-aware systems make recommendations, they incorporate contextual factors, such as time and location, while multi-criteria-based approaches utilize a broader range of evaluative criteria and offering more relevant suggestions. For making recommendations, contextual information and multi-criteria ratings have not been used together, despite both approaches having advantages in producing more accurate and personalized referrals. To fill this gap, a context-aware, multi-criteria recommender system is needed. We employ a deep learning model that integrates context-aware recommendation systems with multi-criteria recommendation systems, instead of processing each separately. Hence the system becomes more adaptive and delivers more accurate recommendations, as contextual and criteria-specific insights are combined to create personalized recommendations. The proposed method provides more accurate predictions than state-of-the-art recommendation techniques when applied to TripAdvisor and ITM-Rec datasets.