“Wisdom of the Crowd” as Personalized Music Recommendation Model for Langit Musik Service

Abstract

We have been witnessing digital music industry has been grown rapidly in recent years. One Innovation which helps this industry improving its customer satisfaction, customer loyalty, and customer engagement, is personalized recommendation system. It has not only been an interesting innovation in digital music service, but also in almost all digital services. Telkomsel, the leader of the cellular industry in Indonesia, would also like to compete in providing“On Demand Music Streaming Service” by launching its own brand, Langit Musik and personalized recommendation system is supposed to be one of the improvements which can be implemented on top of it. Unfortunately, this personalized recommendation was not yet implemented. This research builds model to predict customer preferences for artists in Langit Musik service, to provide more personalized recommendations for each customer. This study applies an implicit preference from the amount of music listening for customers in period of l and3 months, from Mobile Apps and Unstructured Supplementary Service Data (USSD). The modeling uses a collaborative filtering approach with matrix factorization method and measure the model accuracy using Receiver Operating Characteristic / Area Under the Curve (ROC/AUC). The AUC value indicates the prediction quality of the model above prediction from random method. In addition, it was also concluded that the matrix factorization method provides advantages in resource efficiency. The contribution of this research is to improve the customer experience, satisfaction, loyalty and engagement to Langit Musik.