KIT | KIT-Bibliothek | Impressum

Interest-based Recommendation in Academic Networks using Social Network Analysis

Toreini, Peyman; Chatti, Mohamed Amine; Thues, Hendrik; Schroeder, Ulrik

Abstract (englisch): Recommender systems are essential to overcome the information overload problem in professional learning environments. In this paper, we investigate interest-based recommendation in academic networks using social network analytics (SNA) methods. We implemented 21 different recommendation approaches based on traditional Collaborative Filtering (CF), Singular value Decomposition (SVD)-based RS, Trust-based CF, and SNA-based techniques for recommending new collaborators and research topics to the researchers. The evaluation results show that SNA-based recommendation outperforms traditional CF methods in terms of coverage and thus can provide an effective solution to the sparsity and cold start problems in recommender systems.

Zugehörige Institution(en) am KIT Institut für Informationswirtschaft und Marketing (IISM)
Publikationstyp Proceedingsbeitrag
Jahr 2016
Sprache Englisch
Identifikator ISBN: 978-3-88579-656-5
URN: urn:nbn:de:swb:90-597657
KITopen ID: 1000059765
Erschienen in GI Edition Proceedings Band 262 DeLFI 2016 – Die 14. E-Learning Fachtagung Informatik : 11.-14. September 2016 Potsdam. [Hrsg.] Ulrike Lucke
Verlag Köllen, Bonn
Seiten 23-35
Schlagworte Recommender Systems, Social Network Analysis, Academic Networks, Collaborative Filtering, Professional Learning
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft KITopen Landing Page