Deborah Gage
SemanticWeb.com Contributor

Released by the Digital Enterprise Research Institute (DERI) in Galway, Ireland, the software combines a search engine with a browser, a mashup generator and an Application Programming Interface for developers. Search results can also be turned into widgets and will remain live even after they're embedded in e-mails or Tweets or blogs.

How well Sig.ma will work remains to be seen. Its results don't always make sense, although they are attractively displayed. The software aggregates links and pictures about a search term, organizes them into facts and lists them in a column - the "sigma" - down the center of a page.

If you're searching for a person, Sig.ma may return his or her e-mail address, the place where he works or his associates. Mouse over any of these facts and Sig.ma will tell you where they came from - it returns as many as 20 sources and lists them down the right side of the page next to the facts.

Several big Web sites - Google, Yahoo, LinkedIn and some others - have started tagging some of their content according to emerging metadata standards so it can be understood by machines, and therefore read more easily by humans.

LinkedIn, for instance, marks first and last names and other basic personal information on its public profiles. Google in May started rolling out "rich snippets" [http://googlewebmastercentral.blogspot.com/2009/05/introducing-rich-snippets.html] - short summaries that accompany Google search results for people or reviews. They show up when Webmasters follow Google's instructions for marking their content with microformat or RDFa tags, two of the better known standards.

But it's still difficult to collect and see semantically tagged information when it's scattered around the Web. The creators of Sig.ma believe they have solved that problem.

"When we first saw the B&W pictures...pop up automatically the first time we ran Sigma we were really excited," wrote Giovanni Tummarello in a blog post announcing the software. "That DERI data had been there forever yet never meaningfully used or integrated -- let alone automatically!

"...But here it was! That file was there, discovered automatically and contributing marvelously to the mashup providing information about papers, (including technical reports that would not be listed otherwise) an extra picture, the phone number, a confirmation of the personal homepage, research projects and more."

Sig.ma's results are skewed by how often information about a search term is semantically tagged - not often for some people. Sig.ma's search for Stefan Decker, a professor and director at DERI, delivered 50 sources, almost all of them right. But the name Deborah Gage was more of a mystery. Sig.ma delivered 14 sources, all but one of them wrong.

Fortunately, however, Sig.ma can be taught - users can accept or reject sources based on whether the facts the software lists about their search terms are true. Sig.ma is hampered by today's Web, Tummarello wrote, where until recently data was marked up semantically "on a best effort-hacker enthusiastic-leap of faith way.

"Now that Google and Yahoo are starting to recognize the value of page markup, it is realistic to expect improvements in data coverage and quality," he said.

Sig.ma is open source, and its creators plan to release the software's index by the end of the week. The reasoning engine will follow.