ABSTRACT
Recommendation systems (RSs) are frequently used to provide suggestions to users based on their preferences. With so much information on the Internet, RS have proven beneficial for combating information explosions and alleviating multiple over-choice difficulties. Various applications, such as e-commerce, medical, transportation, agribusiness, and entertainment, have welcomed RSs as automated online assistants. The trick is how quickly the recommendations are provided; otherwise, the opportunity may be missed. This chapter presents a near-real-time RS based on utility itemset mining by employing the highly efficient EAHUIM (Enhanced Absolute High-Utility Itemset Mining) algorithm, which has been proposed for finding the associations between items in a database. Also, as the real-world data is dynamic, the problem of transaction-stream fluctuation has been considered, and an adaptive load strategy has been proposed for the model. Various experiments on a real-world dataset result in near-real-time and customized recommendations, demonstrating the model’s efficacy.
