Prediction of user interest based on collaborative filtering for personalized academic recommendation

The development of Internet provides academic researchers with abundant information, which also brings them heavier and heavier information burden, because to obtain what they exactly want from the huge amount of resources will greatly affects the efficiency of information seeking. To alleviate user...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2012 2nd International Conference on Computer Science and Network Technology S. 584 - 588
Hauptverfasser: Yu, Jie, Xie, Kege, Zhao, Haihong, Liu, Fangfang
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 01.12.2012
Schlagworte:
Accuracy
Collaboration
Collaborative filtering
collaborative user
Data models
History
information quantity
Internet
key concept vector
key relation matrix
Predictive models
Semantics
user interest
Vectors
ISBN:1467329630, 9781467329637
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The development of Internet provides academic researchers with abundant information, which also brings them heavier and heavier information burden, because to obtain what they exactly want from the huge amount of resources will greatly affects the efficiency of information seeking. To alleviate user's information burden, academic recommendation which aims at automatically providing articles to researcher according to their interests has caught much attention. In academic recommendation, how to accurately predict user interest is a key issue. This paper presents a prediction method based on collaborative filtering. First, it is implemented based on refined user profile in which concept weights is adjusted. In the adjustment, the continuity feature of user's browsing content is taken into account, which is helpful in discovering collaborative users. Secondly, key concepts are selected from refined user profile. And collaborative users are discovered based only on key concepts, which can improve the efficiency of prediction. Thirdly, when extracting concepts that can represent user future interest, information quantity is presented as the evaluation attribute. In addition, semantic relations between concepts are considered when computing information quantity, which can ensure the accuracy of prediction. Experimental results demonstrate the validity and effectiveness of this method.
ISBN:1467329630
9781467329637
DOI:10.1109/ICCSNT.2012.6526005