Just rank it: the Synapsify approach to analyzing data

Stephen Candelmo was a lawyer in Washington D.C. in 2008 when a colleague introduced him to a client named Lawrence Au and Au’s technology for analyzing English-language text. Au showed off how it was possible to analyze a book and bring to the fore short passages that best sum up its contents. Candelmo was intrigued.

“Does this rank reviews?” Candlemo recalls asking Au. Candelmo felt that consumers are overwhelmed with reviews to consider and that we lack a tool to find the most credible and meaningful ones.

“I’m not sure. We’ll see,” said Au, who has done work for the Defense Advanced Research Projects Agency (DARPA).

About three years later, Au got back in touch with Candelmo.

“Remember that question you asked me?” Au said to Candelmo. Candelmo did. “Yeah, it can,” Au said.

Structure_LaunchpadBy that time, Candelmo had done work on a startup focused on consumer reviews, Klaggle, but it wasn’t working out. So Candelmo and Au started talking about how to applying Au’s technology to review analysis.

They got admitted to The Fort, a startup accelerator Fortify Ventures runs in Washington, D.C. After graduating, they decided to focus on analyzing text of all kinds, not just reviews, and they started a new company, Synapsify, which has now taken on $750,000 in venture and angel funding.

Rather than ingest information and give back newly generated summary text in response, the way Narrative Science does, or show graph relationships the way Ayasdi does, Synapsify evaluates any type of text and ranks statements in terms of resonance or importance. It’s an improtant problem, which is why the company is a Structure LaunchPad finalist. See their pitch and meet the founders at our Structure conference in San Francisco on June 19-20.

The Synapsify tool has clear business applications, although it can also, say, examine song lyrics and come up with the most important lines. This example resonates strongly with Candelmo. Last year he pulled up the lyrics to one of his favorite songs by Bruce Springsteen, “Thunder Road,” and threw the lyrics at the Synapsify engine. The lines that floated to the top were “To trade in these wings for some wheels.” “It was amazing — a little scary,” he said. “… It was right on.” Those were the lyrics that resonated most with him, too.

Based in Bethesda, Md., Synapsify currently offers service to a few unnamed pilot users through an application programming interface (API). One user has gone from spending up to eight hours evaluating Facebook (s fb) posts that mention its big-brand clients to doing the same thing with Synapsify in just a few minutes, Candelmo said. “Self-serve apps” are now in the works.

To succeed, Synapsify must stand out next to other text-analytics players. Lexalytics and Lymbix are just a couple of startups the company has to contend with. What’s more, to scale, Synapsify needs to add support for languages other than English. For now, it appears to have strong foundational technology and deeply-held evangelism that could convince companies of the product’s value.