Embedding Model Design for Producing Book Recommendation

Internet services often recommend contents to users in order to maintain the interaction. Recommendation system plays a major role to formulate and produce a series of recommendation based on users' behavior. Surprisingly, user-generated scoring or known as ratings are the main raw materials to learn the pattern of favorable contents of each user. As a part of collaborative filtering strategy, rating is considerable to be included in formulating recommended contents. In this research, the basic formulation to recommend books is discovered. The recommendation system has been tested on one random-picked user behavior and successfully has generated five recommended books by analyzing prior activities. The embedding model produced the recommended books with 59% of accuracy. This research is done to provide an insightful experience for developing content recommendation system in Binus University's corporate learning system.