Evaluating and Implementing Collaborative Filtering Systems Using Apache Mahout

Abstract

Collaborative filtering is a successful approach where data analysis and querying can be done interactively. In large systems that contain huge data or many users, collaboration is often delayed by unrealistic runtimes. In any electronic application, the recommender systems play an important role as they help in making proper decisions on the basis of the recommendations that the system provides. Today there has been a dramatic increase in the amount of online content. Recommender system software's help users to navigate through this increased content that is collected from users. A recommender system helps a user to make decisions by predicting their preferences, during shopping, searching, or simply browsing, based on the user's past preferences as well as the preferences of other users. In this paper, we explore different recommender system algorithms such as User- Collaborative and Item-Collaborative filtering using the open source library Apache Mahout. We simulate recommendation system environments in order to evaluate the behavior of these collaborative filtering algorithms, with a focus on recommendation quality and time performance.