CS224W Project: Recommendation System Models in Product Rating Predictions

A product recommender system based on product-review information and metadata history was implemented in our project. The primary goal for our recommender system is predicting the rating value that a user will give to a product. We used collaborative filtering model with both user-based and itembased strategies, matrix factorization model and a graph-based Network Inference model as our rating prediction models. We evaluted the performance of these models on Amazon Product co-purchasing Network metadata Dataset. We also discussed the advantages and weakness of them.