Integrating Situation Awareness into Intelligent Recommender Systems

Recommender systems play a key role in managing information overload by adapting suggestions to user preferences and contextual factors. However, traditional ContextAware Recommender Systems often treat context as a static collection of descriptors, without interpreting what such contextual cues actually mean in relation to user goals, leading to limited understanding of user intentions and reduced adaptability of the generated recommendations. This paper introduces a Situation-Aware Approach grounded in Situation Awareness theory and Situation Awareness Oriented Design, which structures the recommendation process into three levels: perception, comprehension, and projection. Particular attention is given in this study to the comprehension phase, operationalized through the mapping f(C) → S, that transforms contextual features into explicit situational representations. Experiments on the MIND dataset show consistent improvements in ranking metrics, along with enhanced robustness and interpretability compared to the baseline FM. These results show that explicitly modeling comprehension realizes more stable and human-like reasoning in recommendation, offering a structured pathway toward more robust and interpretable recommender systems.