Online Product Recommendation System by Using Eye Gaze Data

Abstract

Recommendation system takes information related to the users' habits or interest or profile to suggest users with more convenient or similar materials that the users might be interested in. In general, these systems mostly rely on explicit feedback techniques (rating, search history etc.) to recommend products. In this case, users need to interact directly with the system. We seek implicit methods (indirect interaction) to relate users' preferences and recommend desired products automatically on the interface in order to minimize the meddling interaction and workload. In this paper, we present a recommendation system that will use users' eye gaze data to apprehend their interest to recommend products as an implicit feedback technique. Eye gaze data can provide information about the products that the users are interested in, without any direct interaction with the system. Eye gaze features is very effective as an implicit interaction technique. So, integrating eye gaze features with recommendation system will generate more user-oriented results. This system will collect users' eye gaze data during an e-commerce website navigation through a web-cam based eye tracker. Other features of a product (ratings, number of orders) were also included in generating the results in order to get more convenient results. Finally, a clustering-based machine learning algorithm was used to group the similar product based on the input data and recommend similar products to the users, implicitly expressed greater interest. In this study, we concluded that users can find their desired products with less physical assertion and more satisfaction.