Enhancement of collaborative filtering using myers-briggs type indicator (mbti) applied in recommendation system

Abstract

Collaborative filtering is one of the most popular recommender systems being used today. Collaborative filtering algorithm depends on the association of one client's activity with another client's activity to discover his nearest neighbors. Related items are expected to be recommended according to his neighbor's similar interests or inclinations. Collaborative filtering algorithm deals with a major problem called the new user challenge or also known as the ‘cold-start’ problem that arises due to the lack of enough information about the new-coming user. The authors have employed an enhanced collaborative filtering algorithm by incorporating Myers-Briggs Type Indicator or MBTI. With the means of identifying each users MBTI personality types to create neighbourhoods, the researchers have alleviated the problem on the lack of similarities between inexperienced users to existing users. In addition, the system can predict new user ratings for each item using the average rating of users in the same neighbourhood. After predicting the rating, the item with the highest rating is recommended to inexperienced users, which provides a solution to the ‘cold-start’ problem