| Autor: |
Géczy, Peter, Izumi, Noriaki, Akaho, Shotaro, Hasida, Kôiti |
| Zdroj: |
Data Mining: Special Issue in Annals of Information Systems; 2010, p317-334, 18p |
| Abstrakt: |
We present a novel recommendation algorithm for browsing assistance systems. The algorithm efficiently utilizes a priori knowledge of human interactions in electronic environments. The human interactions are segmented according to the temporal dynamics. Larger behavioral segments – sessions – are divided into smaller segments – subsequences. The observations indicate that users΄ attention is primarily focused on the starting and the ending points of subsequences. The presented algorithm offers recommendations at these essential navigation points. The recommendation set comprises of suitably selected desirable targets of the observed subsequences and the consecutive initial navigation points. The algorithm has been evaluated on a real-world data of a large-scale organizational intranet portal. The portal has extensive number of resources, significant traffic, and large knowledge worker user base. The experimental results indicate satisfactory performance. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
