The Discovery Channel: Getting Content Discovered

Page 2 of 3

Search For Community
Barry Smyth, head of the department of computer science at University College Dublin, believes search can do more than link people with content. He thinks that linking actual users through their search patterns and information preferences could create communities of practice since all members would be committed to seeking the same answers and solving the same problems.

Smyth's approach is to "capitalize on the observation that many searches do not exist in isolation." More often than not, he stresses, searches can be related to a community of like-minded searchers. "For example, within my work environment I am likely to be searching for information similar to many of my colleagues; so my work colleagues form an ad hoc search community." To tap this creative energy and ultimately create communities of practice, his prototype I-SPY technology works in collaboration with existing search engines "to post-process search results and ensure the right community members get the right results," Smyth explains. It also learns from community search habits to solve the legendary vague query problem. Put simply, it watches search patterns and figures out that a community of biologists is interested in animals and not cars when its members search for the term jaguar, for example.

"This approach can significantly improve the quality of a search engine's results by ensuring that community-relevant results are promoted to top positions within the result lists, even for the vaguest of queries," Smyth says. "This can dramatically improve mobile search because it guarantees that relevant results will appear within the top three results where the community can find it." What's more, the technology provides fertile ground for serendipity, enabling users to find something unexpected and useful while searching for something else entirely. 

Smyth's focus is mobile search, a space he thinks is more challenging and potentially more rewarding than general online search. Screen size, battery-life, and text-input issues abound, making it necessary to rethink content search and discovery from the ground up, he argues. On the upside, the capability of mobile devices to deliver location-based services opens the door to a slew of new services and search scenarios. "There is the possibility to not only link users in communities; there is also the technology present to someday introduce them to each other (if they're in the same vicinity). 

"In the past, if users wanted to make themselves heard on the Internet they usually set up a Web page," Smyth says. "What we're saying now is there's something very different happening. Instead of setting up a Web page, there's a community that users belong to based on their search behavior. We can give that community visibility and other users can discover it." Smyth's community-based search does not maintain profiles for individual users or track the behavior of individuals. Instead, only a community profile is maintained so that it is not possible to trace individual search behavior. 

Discovering information is important, but even the most relevant search results can be overwhelming. Combine that problem with an explosion of information sources including blogs and a host of obscure yet influential Web sites, and online search can become an exercise in frustration.

The problem is particularly acute for companies using search tools to sift through millions of pages of information to discover emerging opportunities and threats. These users require tools that spot trends and transform reams of data into actionable information. They also increasingly require technology that will allow them to listen in on (and make sense of) the infinite consumer discussions taking place online. As the authors of the milestone business book, the Cluetrain Manifesto, point out, "markets are conversations." 

Against this backdrop, companies worldwide face an "adapt or die" dilemma. They either have an ear to the ground or wind up with their backs to the wall. Take the example of Kryptonite, maker of bicycle locks: When bloggers posted they could pick the lock with a Bic ballpoint pen, the company failed to pick up on the buzz in the blogosphere, triggering a major problem. Ingersoll-Rand, Kryptonite's owner, has publicly said this mistake may cost as much as $10 million, a big chunk of Kryptonite's estimated $25 million in revenues. 

This disaster might well have been avoided, if the company possessed the technology to identify and visualize trends affecting its brand assets, argues Factiva, a Dow Jones & Reuters Company. To help companies discern information from noise, the company offers Factiva Insight: Reputation Intelligence, a tool that allows users to monitor known issues and discover emerging ones across mainstream media and user-generated content such as blogs and message boards. 

"The idea is to provide users with tools that enable them to spend more time analyzing and less time searching," observes Dennis Cahill, Factiva product VP. "Essentially, we're giving customers a security blanket across the some 700,000 documents published every day, allowing them to track all of the trends and issues they know about, but also to discover the issues they don't know about using the same technology." 

In January, Factiva re-launched its flagship product with a new interface that will "take some of the Factiva Discovery technology and make it part of the normal search interface for Factiva," Cahill says. Customers who use the Factiva search box will receive more than results; they'll be presented with "powerful visualization around discovered companies and discovered news within that results set." What's more, the technology will provide users dynamic clustering of the results, allowing them to better understand the key themes that exist within the overall results set.

Factiva also has plans to expand its capabilities to recommend information to users on the basis of their search and reading habits. "We call it implicit personalization," Cahill explains. If a user always reads a specific source, then the system might ask if the user would like the content delivered automatically when it's published. "As individuals use our system, they are, by definition, giving us information about their job function and the tasks they perform on an everyday basis," he continues. "For us to be effective at that next level of value-creation and making our customers even more efficient, we need to understand what they're doing and configure the system along the path that they normally execute."

Page 2 of 3