International Journal of Scientific Research and Engineering Development

OTT platforms overwhelm their users with a plethora of multimedia content, and it is difficult to figure out what they really want to consume. Conventional recommender systems, which are mostly based on historical data and collaborative filtering, often fail to capture user intent in the current session, such as understanding what device is currently in use, where the user is, and how the session is progressing. In this paper, a hybrid framework for a privacy-preserving, session-based recommender system is proposed, where personalization is viewed as a real-time, context-aware process, unlike conventional user profiling-based approaches. This framework leverages K-Means clustering and Swarm Intelligence optimization to finetune the ranking of recommendations in real time. One of the key contributions of this paper is a dynamic session modeling technique, where anonymized user IDs are generated based on device fingerprinting, timestamps, and network-related parameters, so that personalization is possible without tracking individual users. Experiments with this framework show that it outperforms traditional methods in terms of precision, recall, and F1-score, while improving diversity and flexibility of recommendations. In other words, this paper proposes a framework for a future generation of recommendation systems for OTTs, which is both efficient and privacy-aware.

Omkar Gilbile, Satish Gujar,"A Privacy-Preserving Session-Centric Hybrid Recommendation Framework for Large-Scale OTT" International Journal of Scientific Research and Engineering Development, V9(2): Page(2307-2319) Mar-Apr 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.