BACKGROUND:  This study is motivated by the question of how are pages displayed in response to Web searches, primarily on Google? Web sites returned at the top of the list in Google searches have greater commercial appeal. How do web results get to the top of the list?

Google takes into account over 100 factors in determining the final ranking of a page. The core of their ranking algorithm is based on a metric called PageRank. PageRank is a link popularity metric, where is a page is considered important if it is link to by many other pages on the Web. Roughly speaking, Google puts a page at the top of a search result, when the page is linked to more frequently than other candidate pages.

The problem is that PageRank is based on the current popularity of a page. If a Web site is popular, it tends to stay popular. What about new pages, which can have high quality content, as determined by users who store that link, but are not linked to by other web sites since it is relatively new?

INNOVATION:  This innovation provides an algorithm to provide a notion of fairness and quality for new Web pages, so that they can have a better opportunity to be highly ranked. This innovation provides a formal definition of Page Quality and a quality estimator, which is a quantitative measurement of page quality. Roughly speaking, page quality is the probability that a user will create a link to a page having accessed it.

DEVELOPMENT TO DATE:  An algorithm to measure page quality was written and experiments made on the Internet. 154 Web sites were measured with data on 4.5 to 5.0 million pages. Interestingly, our estimated quality predicted the future PageRank more accurately than PageRank itself. Thus the results justified our quality estimator.

Reference: UCLA Case No. 2004-183

Published Patent App.: 2006/0294124