Web Link and Traffic Analysis for Recommendation

Brian D. Davison

Abstract (1 pages)
PDF

First paragraph
The Web has two pervasive sources of activity that can be useful for recommendation systems. One is the traffic on the Web, reflected in individual user histories as well as proxy and origin server usage logs. The second is in hypertext link authoring. Both provide information that can often be treated as a value judgment. A user who clicks on a link, in effect, believes that there is some content of value at the destination. The significance of link creation is thus even stronger, as it reflects the judgment of the author (presumably) after viewing the target content and assessing it to be of sufficient value to spend the time to make the connection available to readers.

In Recommender Systems: Papers and Notes from the 2001 Workshop, Co-located with SIGIR 2001, New Orleans, LA, September 2001.

Back to Brian Davison's publications


Last modified: 7 November 2021 Brian D. Davison