TEKNIK POLLING DI RECOMMENDER SYSTEM COLLABORATIVE FILTERING UNTUK PEMBELAJARAN DARING

Abstract

Currently, the Recommender System (RS) is a method that is widely used to help sort out information, which is currently very large. Without a Recommender System it will be very difficult to sort out the information one by one as needed. Sorting information in a RS is not the same as searching for information, as we do a search for files on storage media by simply writing a few keywords to find the files needed. RS sorting is by looking at the magnitude of a value obtained from drawing conclusions after analyzing the available data, either the user data itself or other user data. Information separation in online learning is also very much needed. Because online learning will be more effective if learners can be provided with the right material. Online learning that is currently available generally provides learning material content such as textbooks in hardcopy-like form. In this study, the online learning system was added with RS technology in order to help students choose the material that the user or students should study so that they can achieve the expected learning objectives. The method in the Recommender System that is widely researched in online learning is Collaborative Filtering. RS with collaborative filtering in order to provide accurate recommendations requires large amounts of data. Big data raises a problem, namely the spread of data occurs in many locations so that it requires complex computation in providing recommendations. To overcome the computational complexity, this paper will discuss the polling technique as a novelty in this study. The research shows that there is an increase in recommendation precision by 20%, when compared to data without polling.
 Keywords: Recommender System, Collaborative Filtering, Polling