The School of Computer Science is pleased to present...
Towards a scalable recommender system framework for sparse data
PhD Dissertation Proposal by: Bahareh Rahmatikargar
Date: Wednesday, 29 May 2024
Time: 10:30 AM
Location: Essex Hall, Room 122
Abstract:
Recommender systems are pivotal in managing information overload and enhancing user satisfaction through personalized recommendations. However, the persistent challenge of data sparsity decreases their accuracy and efficacy. To address this issue, we propose a novel scalable framework that aims to improve recommendation accuracy and user satisfaction. Challenges such as balancing quantity and quality, domain-specific understanding, and scalability management are inherent in addressing data sparsity. As systems scale, the complexity of efficiently handling data sparsity grows, necessitating robust algorithms and infrastructure. Leveraging innovative techniques such as anomaly detection in single-domain recommendation systems and community detection and alignment techniques in cross-domain scenarios, our framework adapts recommendation strategies to address these challenges. We hypothesize that our scalable recommender system can enhance recommendation accuracy across diverse scenarios, as evidenced by relevant accuracy metrics such as Recall and MRR.
Keywords: Recommender Systems, Data Sparsity, Anomaly Detection, Community Detection
Thesis Committee:
Internal Reader: Dr. Saeed Samet
Internal Reader: Dr. Dan Wu
External Reader: Dr. Mitra Mirhassani
Advisor(s): Dr. Ziad Kobti and Dr. Pooya Moradian Zadeh
