Hyper-Contextual Search Results with Swicki

Eurekster’s Swicki search service officially launches later today (November 16, 2005).

Eurekster, a twenty person company located in San Francisco and New Zealand, has a profitable business (called Search Publisher) that provides customized search results to a number of large websites. Steven Marder, Eurekster’s CEO, tells me Eurekster’s current products are generating 25+ million monthly searches.

With Swickis, they’ve taken the basic technology and added on a “do it yourself” interface to allow a much larger number of sites, particularly blogs, to also integrate search direclty into their content.

Swicki’s are “community powered” in a sense and their website focuses on this.

What is a swicki?

A swicki is new kind of search engine that allows anyone to create deep, focused searches on topics you care about. Unlike other search engines, you and your community have total control over the results and it uses the wisdom of crowds to improve search results. This search engine, or swicki, can be published on your site. Your swicki presents search results that you’re interested in, pulls in new relevant information as it is indexed, and organizes everything for you in a neat little customizable widget you can put on your web site or blog, complete with its very own buzz cloud that constantly updates to show you what are hot search terms in your community.

And certainly the community has a role in creating more relevant results. But what really interests me about their technology is that the tweaking by the publisher along with community actions combine to create extremely contextual, or hyper-contextual (my words, not theirs), search results.

Two somewhat different examples of live Swickis (they’ve been in private beta for a while): see our Web 2.0 Workgroup (scroll down a bit) and Jeff Clavier‘s right sidebar.

Swicki’s are based on Yahoo’s search API for base results. The publisher customizes the search engine by adding keywords that are always added to the search results. And, in a similar way to Rollyo, Swickis allow for the publisher to name specific websites that have content relevant to the search. For instance, a gaming site may include other gaming website URLs as important, and Swicki will put results from those sites on the top. A publisher can also block results from certain blogs. For our Workgroup, we selected all participating site URLs as the most relevant content.

Swicki’s also have a related “buzzcloud”, which is a tagcloud of commonly searched terms. The buzzcould can also be edited by the publisher to add or remove terms, and a spam filter disallows any single IP from influencing the buzzcloud results too heavily.

Swickis are completely free. They will soon have contextual advertising served along with normal search results if the publisher chooses, but there will be no penalty or fee if they publisher wishes to keep advertising out. They are also adding analytics to allow publishers to monitor search statistics.

A press release should be out soon and I’ll link to it once it’s available. Check out their blog as well for a post.

Brian Benzinger posted a long review of Swickis last month when they first went into private beta. Thanks Susan for introducing me to the company.