hotpages.pl

hotpages.pl is a perl program that will generate HTML for a table of links on a website that are hot.

In this case, "hot" means not only that the pages are being hit a lot, but also that the pages are, in some fashion, hard to get to at that site. Fow instance, a page may be 10 links deep from the main page, so that page will be harder to get to than another page which is only 2 links deep.

hotpages.pl determines which pages are hot based on an algorithm described in "Adaptive Sites: Automatically Learning from User Access Patterns" by M. Perkowitz and O. Etzioni, Proceedings of the Sixth Int. WWW Conference, 1997.


Generally, hotpages.pl is run once a day, usually at night, and it will crank away on your website's access log and decide which pages deserve to be in the hotpages table.

To see an example of the output of hotpages.pl, here is an (inactive) sample home page to a small site of about 400 pages which represent a user manual for the C++ Standard Template Library. These pages were originally written by Silicon Graphics Computer Systems, Inc., but I modifed the top level index.html page to include the output from hotpages.pl which creates a small table at the bottom of the page. Within this table are the Top 5 hot pages for this website, along with the scores computed by hotpages.pl (these scores are not really necessary, but included in the table for illustrative purposes).


You can get the perl source (zipped) or (gzipped).


Back to Programs
Back to Dan Pape's Homepage