By Hays Goodman
Associate Editor

Imagine retrieving an article from a newspaper Web site about a particular state’s budget shortfall and how they plan to cope with it. The article is written focusing on one legislator, and you are curious about seeing how other elected representatives would deal with the issue.

Current technology would most likely force you to go to the paper’s archive search feature, where you would try to type in search terms that would make the proper match. It is more than likely that unwanted results would be returned as well, interspersed with the information you were actually looking for.

Now suppose instead of that cumbersome retrieval process, you were able to simply hit a Related Pages button, which would call up stories that were conceptually related to the one you’re reading. That’s the vision behind a small company based in northern California, Active Navigation Ltd., and their flagship product called Portal Maximizer.

The company was founded in 1994 in Southampton, England, and initially went by the name Multicosm. The initial purpose in the company was to commercialize research that was generated by the Multimedia Research Group, which is now known as Intelligence Agents Multimedia Group of Southampton, United Kingdom.

The core technology behind the company’s products is the ability to insert active hyperlinks on-the-fly in almost any textual format document. Microcosm was the first release, and two early purchasers were IBM and the U.S. Navy. The next generation, Webcosm, was introduced in 1998 and still forms the core of the technology used today, which is called Portal Maximizer and was released in early 2001.

Active Navigation is able to generate pop-up windows containing related articles, using intelligent concept relationships that are contained in "linkbases."
Photo courtesy Active Navigation

Portal Maximizer is implemented essentially as a Web proxy server. When the user requests a Web page, the browser will be directed to the Webcosm proxy. Webcosm will fetch the page from the original location and annotate the page with extra links before passing the modified Web page back to the user’s browser. When the webmaster has activated this feature, the user will see portions of the text transformed into hyperlinks, which are derived from what is known as a linkbase. This linkbase contains at a minimum a source word or phrase, a destination URL and a description of the link. The linkbase is generated automatically by crawling the Web site at predetermined intervals, with the results fully tunable so that by moving a slider one can decide how broad or narrow particular themes can be. By making the themes broader, nearly every word in a document could theoretically be hyperlinked, but by selectively tuning that variable, more relevant results are obtained. Multiple linkbases can be used, so that different groups of users could see different results, depending on their profile or interests.

Updates to the product in 2001 brought out several new features. Chief among those are Related Pages and Summarization. Related Pages allows the inclusion of a JavaScript on the page that will pop open a smaller sub-window when a button is clicked on-screen, showing stories that are related, given the linkbase that is active at the time. This is useful in the case where the webmaster would rather not provide active links within the actual stories or document. Summarization selects the most important concepts from a story and presents them as a condensed form.

Jane’s Information Group, a well-known U.K.-based publisher for the defense, aerospace and transportation sectors, implemented Portal Maximizer on their collection of 20 Web sites, using the technology to index, categorize and centralize access to the extensive network.

“The system has dramatically improved the navigation on our site,” said Simon Mitchell, managing editor at Jane’s. “Not only are we saving customers time in finding the information they need, but we are also opening up new pathways to content based on how they navigate the site.” Jane’s also decided to use the software to promote value-added, subscription-based content services that would actually correspond to an individual user’s navigational preferences.

“We soon realized that the software was eminently more powerful than we imagined. We went from developing a very specific application to a system-wide integration that completely revamped the way we market our product to site visitors.”

The Seattle Post-Intelligencer, a 170,000-circulation daily, is currently in testing with Active Navigation and will be the first U.S. newspaper to roll out the service. Seattlepi.com has had a Web presence since July 1996 and they have averaged 375,000 unique users and 10.8 million page views per month over the last six months. Lee Rozen, general manager, sees benefits to the reader from Active Navigation.

“We hope users will view more stories on each visit, because they can find more stories that interest them very easily. There are around 100,000 stories active on the site at the moment, most of them unlinked. We’ve found the Related Stories functionality cuts down on the number of times customers have to use the search function,” Rozen said.

The paper has found the most challenging part of the installation to be tuning the linkbases. “Generally, making the technology work on our site has not been a big deal … but then we aren’t live yet, either. We won’t go live until we have satisfactory ‘related story’ results at least 80 percent of the time.”

Carl Swanson, vice president of sales for the North American division of Active Navigation, sees benefits for the newspaper industry using this technology.

“Most of the rich historical information owned by many newspapers is virtually unattainable by everyday users. Many of the people we’re working with believe, and we concur, that there is revenue potential in those untapped hidden assets.”