Sistem Rekomendasi Content-based Filtering Menggunakan TF-IDF Vector Similarity Untuk Rekomendasi Artikel Berita

Abstract

The population of active students in the Informatics Bachelor Program, Universitas Amikom Yogyakarta, in the odd semester of 2021 is 3,870. Efforts to track interest in the three concentration options were carried out early on through article literacy recommendations. Various articles are produced continuously and provided on an ongoing basis to students. However, the many articles offered daily make students overwhelmed and tend to choose articles that do not match what they want. To help solve this problem, recommender system is developed. A recommender system helps to estimate the prediction value or relevancy of an article and create a ranking according to user's taste. Content-based Filtering technique is used in this research. Using the dataset from Kabar Informatika news portal of University of Amikom Yogyakarta, the developed Content-based Filtering Recommendation System is able to produce Recall@5 score at around 73% and Recall@10 at around 80%.

Enhancing Personalized Recommendations: A Study on the Efficacy of Multi-Task Learning and Feature Integration

“…They suffer, however, from the cold-start problem and data sparsity issues [18,19]. Content-based filtering, on the other hand, is increasingly leveraging advanced techniques for similarity computations based on user profiles and item metadata [20][21][22], which can help overcome some limitations of CF, but is still dependent on the quality and availability of data [23].…”