AliExpress-A Collaborative Recommendation Algorithm

The objective of using Recommender systems in e-commerce is to guide or assist a customer in his purchase by providing personalized suggestions through the large product assortments, and thus helping him in better decision making. We try to investigate the reason for item-based collaborative filtering technique being preferred over popularity-based recommendation system. In order to do so we developed a personalized collaborative recommendation system based on item-based and popularity based techniques. Item-based collaborative filtering technique involved construction of a co-occurrence matrix to determine user-item interactions. Popularity-based recommendation system was developed to determine if popular features played an important role in consumers’ purchasing decisions. From the results obtained, we inferred that item-similarity model provided a better performance as it had higher precision and recall values in comparison to the popularity model.