Valuing a web page

Web pages, as with any other page, are not created equal. Nor is the information they contain.

We may read a newspaper by doing no more than scanning the headlines and reading an article or two. In cases, though, we may read an article of particular interest several times over.

We may make a point of passing on the article to a colleague or significant other. We may tear it out from the newspaper or keep the newspaper because we value the particular story so highly.

It is more less the same in the online environment. We scan hundreds of headlines or links to web pages every day. Some we click and start reading, but do not finish. Others we read in full, and perhaps, re-read.

We save some web pages to our hard drives either in full or just a part and we email the text or link to friends, family and colleagues. In other cases, we print out copies of pages we value highly so that we can easily reference these pages if we want.

There are many pages where we do not read to the bottom of the page, but there are some, that no sooner that we have finished, we post to people we know will find the page of interest.

This email filter has quickly become a way of sorting very high value content from the dross. The best stuff flies around the web as soon as users come across it. It quickly goes viral.

JiQA is premised on the basis that web pages are not created equal. It sorts out which pages are of high-value from the low-value pages while you browse. It does this by time: do you spend a lot or little time on the page relative to the content it contains?

Later releases will include other activity measures listed above, namely whether the page is saved in whole or in part, whether it is emailed and/or printed, but the present, beta release contains a time-based ranking.

Aha, I hear some of you say, what about the page which is opened, but is not read, because the user goes off to make a cup of coffee?

The answer is that only counts closed sessions are counted for ranking purposes. The last page browsed in a session will not be ranked.

A history of browsed pages is maintained by keyword and ranking on JIQA Local on the user’s hard drive. This means that the user can quickly retrieve ranked pages by keyword.

The url of the browsed page, plus the page value, is also communicated anonymously to JiQA Central, where the page value is pooled with that of other users of the JiQA network.

A key value is whether or not a page is clicked in the first case. We scan hundreds, if not thousands of single-paragraph web links each day, a relatively small proportion of which we decide to click.

This, on the face of it, makes JiQA susceptible to click fraud. The more you merrily click a web page, the higher you go up in the rankings. But JiQA has limits on the interaction users can have with JiQA

Central, meaning that you can bring up the same page any number of times, but only one will be counted.

How well does this ranking system work? In truth we believe we will only know the answer once we have an active base of users, whose collective browsing tells us what is normative and what are aberrations to be discarded.