SCIENTIFIC ARTICLES RECOMMENDATION SYSTEM BASED ON USER’S RELATEDNESS USING ITEM-BASED COLLABORATIVE FILTERING METHOD

Scientific article recommendation still remains one of the challenging issues in education, including learning process. Difficulties in finding related articles from research history and research interest have been experienced by students in collage affecting the duration of study and research time. This paper proposed a new solution by building a search engine to collect and to recommend articles related to student research topics. The system combined the web scraping method as an article data retrieval technique on google scholar and item-based collaborative filtering to recommend the article. Parameters result produced based on items of user’s history, including item-searched, clicked, and downloaded. The system was built on a web-based scientific article recommendation system using python programming language. This system recommends articles based on the preferences of users and other users who are affiliated and who have an interest in the same item. This research showed that the validation result from the system obtained a recommendation accuracy value over 0.516801. The percentage of the RMSE error value of the recommendation system is 8.62%, or in other words that the accuracy of the recommendation system is 91.28%.