Collaborative Filtering in an Offline Setting Case Study: Indonesia Retail Business

In the past decade, most modeling efforts to date have been focused on the application of recommender systems in an online setting. However, only a few studies have exclusively addressed the actual challenges that arise from implementing it in an offline system. Although the principles of recommender systems implementation between the online and offline commerce are almost identical to some extent, applying the algorithm in the offline environment has its own unique challenges such as lack of product rating and description. Furthermore, most of the customers in the offline retail tend to purchase favorite products repeatedly in short periods. Overcoming such shortcomings could help offline retail to identify the right product that has a higher likelihood to be purchased by a specific customer, and hence increasing revenue. This paper proposes the use of Item-based Collaborative Filtering algorithm as recommender systems to address the limitations of the offline setting.