A Recommender Ranking System project that uses various models for recommendation and attemps to address the cold-start problem using advanced ranking algorithms.
Recommendation systems have garnered significant attention in recent years, prompting exploration into a variety of AI & ML methodologies to address them.
In this project, we developed traditional collaborative filtering models such as Singular Value Decomposition (SVD) and Alternating Least Squares (ALS) Matrix Factorization. Additionally, we ventured into implementing classical ranking techniques to handle the cold-start issue. These methods encompass PageRank, Elo, Elo+KNN, Massey Rating, and Average Dominance Rating.
Moreover, we explore content-based methods like LightGBM to further enhance our system's capabilities.
Our work can be adopted to the benchmark dataset MovieLens and an interesting speed dating data set from Kaggle.