"Discovery" has become one of the Web 2.5 buzzwords over the last year, though I have to admit the word enjoys the benefit of fuzziness. Nailing down precisely how, where and when people are to "discover" what they need rather than just search for it is a challenge.
The recommendation engines at e-retailers like Amazon can be enormously effective when they are well integrated with the shopping experience and finely tuned to your past browsing and buying habits. But individual Web sites and e-tailers represent a limited universe of content, where a publisher should be able to customize the experience based on a frequent user's habit. But how does one wrangle the much messier, diverse and limitless content of the Web itself? How can the Web be lassoed into becoming a recommendation engine of content in which the user is interested?
One group of academics in British Columbia is pursuing a discovery/recommendation engine of sorts that rides alongside most search engines. The Worio engine (www.worio.com) lets a user search via most of the common engines to deliver the usual results in the left-hand part of the screen. But the right side of the results display renders deeper and different content recommendations that marry the keyword you entered with a much larger profile of your past interests, as well as the buzz around these topics across various social media venues. "It is Web-scale discovery," claims Worio CEO Ali Davar. His team of 10 researchers spun out from the University of British Columbia and landed $3 million in angel funding to pursue a new model of discovery that incorporates social cues in order to determine what the user needs to know. "We are trying to get people the things they are missing," says Davar.
When I first logged into Worio, it asked to use Facebook Connect to link me to that social network and incorporate my profile. It then scanned the profile and my social graph to get a baseline knowledge of my interests and associations that might inform the recommendation engine. The system will learn from my searches and clicks what kinds of information I pursue.
At the same time, though, the real heavy lifting at Worio comes in indexing all of that content in order to marry it more precisely to my needs. The Worio engine looks at the content that is already tagged on the Internet and then uses that information to auto-tag all of the untagged content it finds. According to Davar, the technology drills deeper into the pages to try to understand the interesting ideas that are a part of the content. In the end, it means that two different people can run the same keyword search and get very different recommendations. Running a search on new PC chipsets might give your IT guy a set of recommendations that are filled with tech specs, but an investor might get a wholly different set that focuses on tech company financials.
In addition to personal profiles, Worio is scouring the zeitgeist. "We monitor the social media sites out there and see how people interact with documents," says Mike Klass, CTO. "We know what is getting Tweeted and shared on various sites, and by analyzing that activity, we can determine the topics. We are monitoring the swarm."
While I can't attest yet to the discovery engine's effectiveness (Worio and I are just getting to know each other), it is clear that the right column of a Worio page drills deeper into content than a standard Google results page that spews predictable search-optimized commercial home pages, ancient items, and the ubiquitous Wikpedia entry. "By default, some internal pages are more interesting than the home pages," says Klass. "We try to target things that are more interesting, with a bias towards what is recent and the zeitgeist -- what is being talked about on the Web and what is interesting to the most diverse set of people."
What is most interesting to me about the Worio model is that it runs the two engines, search and discovery, side by side. They work in a complementary fashion and can highlight the strengths and weaknesses of one another. There are times when one does want a Wikipedia entry to float to the top, or we expect now that the most relevant brand did indeed throw the most SEO dollars at their site to ensure they topped the results. But the Worio results often skew more to the content side. They link into relevant articles rather than companies.
Neither Davar nor Klass fantasize of challenging Google with a better form of search. "Keyword search has reached its maturity phase, but discovery is nascent as a business model," says Davar. "The idea is to make the technology as good as possible so that one of the big players gets interested."
How advertising fits into this model is anyone's guess at this point. Clearly, Worio is creating a unique layer of data, both around individual users as well as around the Web at large. Worio hasn't started contemplating yet how this monetizes for marketers. Of course, if all goes according to plan and the company attracts the attention of a major engine, then ultimately this becomes Google or Microsoft's problem to solve, perhaps. One thing we do know for sure. The final integration of a Worio-like discovery engine into search will not take its current place in the right-hand column of the page. After all, we think Google already discovered for itself a more lucrative use for that space.