Title/Titel | A Comparison of Different User-Similarity Measures as |
|---|---|
Author/Autor | Tamara Heck |
Source/Quelle
| Information Science and Social Media – International Conference, Åbo/Turku, Finland, August 24-26,2011 |
Language/ Sprache | English/Englisch |
A Comparison of Different User-Similarity Measures as Basis for Research and Scientific Cooperation
New web-technologies can facilitate scientific and business work and research in many ways. The critical aspect is which structures and methods are to be used to elicit the best out of the existent resources? The information-overload is present in all-day life and sciences. Recommender systems try to solve this problem and have not only established in e-commerce, but also in the collaborative web, such as on Social-Bookmarking platforms [Heck/Peters 10]. In this paper, we analyze a database of records found on Bibsonomy, CiteULike and Connotea – three Social-Bookmarking Systems for scientific references – and explored the tripartite connection of users, documents and tags by three measurement methods. We concentrated on two research questions concerning the recommendation of similar users: 1) Are there differences when we apply different coefficients (Dice, Cosinus, Jaccard-Sneath)?, and 2) Are there differences when we apply shared documents or shared tags?
Tamara Heck
Department of Information Science
Heinrich-Heine-University Düsseldorf
Universitätsstraße 1
40225 Düsseldorf
E-Mail: tamara.heck@uni-duesseldorf.de