Author(s) :

Shubham Sharma .

Volume/Issue :

Abstract :

Recommendation systems are being used a lot in helping users find content they may like, especially on shopping and OTT services. But the problem is, many of these systems are working in certain ways that people cannot see or understand. They often don’t explain how they choose what to show, and sometimes, they are not fair to all users. In this paper, we present a recommendation system built using Graph Neural Networks (GNNs), specifically a Graph Attention Network (GAT). This model helps highlight which past interactions matter the most, making it easier to explain why something is being recommended. At the same time, we apply techniques to reduce bias like making sure that less popular items still have a chance to get recommended.We have tested our model using the data from Amazon Reviews. The results showed our model performs as well as other popular methods/ models while doing a better job of balancing recommendations and providing explanations that are easier to understand. We also talk about the challenges and future directions for building recommendation systems that are smarter, fairer, and more transparent.

No. of Downloads :

0