User session interaction-based recommendation system using various machine learning techniques

A recommendation system can help users to find relevant products or services that they might want to buy or consume. In most of the real-world applications, user's long-term profiles may not exist for a large number of users, which might be the reason that they are visiting the website for the first time or they may not be logged in. The frequent change in user's behavior requires a system which captures the present context or the short time behavior in real time. To predict the short-term interest of a user in an online session is a very relevant problem in practice. In this paper, we have applied eight machine learning models on the different datasets from different domains to check the performance of models and compared the results. From the obtained results, it is observed that the session-based KNN (SKNN) and its variants give promising results compared to the other's methods.