Browse Prior Art Database

A Probabilistic Method for Autonomic Updates of Content for an on-demand Web site

IP.com Disclosure Number: IPCOM000029030D
Original Publication Date: 2004-Jun-11
Included in the Prior Art Database: 2004-Jun-11
Document File: 2 page(s) / 44K

Publishing Venue

IBM

Abstract

Disclosed is a method that allows Web sites (or comparable systems) with changing or evolving content to "test" new content with a limited audience before serving it to the entire requesting population. A Web site, in addition to a single Web page, can have alternate versions of that page with links to new or different content. These alternate pages can be served to a small percentage of the requestors. As the clients click on the link(s) to the new content, the percentage of people who receive the new content can be increased. When a threshold of clicks is reached, then the new content can automatically, via the Web Server, replace the old content (either by merging various alternate pages or by replacing an entire page). Conversely, if the new content is never clicked, then it can be logged as such and discarded as not interesting to the users.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 1 of 2

A Probabilistic Method for Autonomic Updates of Content for an on -demand Web site

Disclosed is a method that allows Web sites (or comparable systems) with changing or evolving content to "test" new content with a limited audience before serving it to the entire requesting population. A Web site, in addition to a single Web page, can have alternate versions of that page with links to new or different content. These alternate pages can be served to a small percentage of the requestors. As the clients click on the link(s) to the new content, the percentage of people who receive the new content can be increased. When a threshold of clicks is reached, then the new content can automatically, via the Web Server, replace the old content (either by merging various alternate pages or by replacing an entire page). Conversely, if the new content is never clicked, then it can be logged as such and discarded as not interesting to the users.

The technology/algorithm for this invention will exist in the Web server, Web proxy, or Web (proxy) cache using the following steps:

Do not present pages with links to new content to all requestors (clients).

Based on a percentage, randomly select a subset of requestors to serve the new pages to. Other requestors will be served the old page. As the new content gains popularity, measured via clicks to that new content

from the page(s) served, serve the page(s) with the new content to an increasing percentage of requestors. When the percentage of requests serviced with the new content exceeds

some threshold, replace the old page with the new page or union of the new pages. If the percentage of requests that are served the new content does not

increase, does not...