Lexalytics Looking Strong as Text Analytics Heats Up for Big Companies, Mobile

7/31/12Follow @gthuang

I wish I could say Lexalytics has analyzed the entirety of the world’s social-media and Web conversations, and deemed it all worthless. Alas.

“That’s not our role,” says Jeff Catlin, the company’s CEO.

Nevertheless it’s a good time to check in with Lexalytics, the Boston- and Amherst, MA-based text analysis company whose technology seeks to understand the meaning and sentiment of conversations on the Web.

In recent years, text and sentiment analysis has graduated from a niche technology used for social-media monitoring and customer relationship management, to a more mainstream feature used by a lot of big companies for business intelligence and other applications. And mobile is the next frontier (see below).

Lexalytics spun out of LightSpeed Software in 2003—to avoid getting shut down—and has grown to become a profitable business, with annual revenue growth of 30 to 50 percent over the past few years, Catlin says. The company has about 20 employees and now counts Oracle and Microsoft among its big customers. Lexalytics sells its software mostly to companies that use its technology under the hood to help other companies and brands understand what people and customers are talking about online.

The technology currently works for English, Spanish, French, Portuguese, and German, and the company has started working on Chinese. (Interestingly, Catlin says software for Japanese didn’t pan out, because the writing style of the culture requires a lot of contextual interpretation by the reader, and Web writers don’t express a lot of sentiment.)

Last year, Lexalytics also helped spin out a company called Semantria, which does cloud-based text analysis. The new startup is a joint venture between Lexalytics, Postindustria (in Los Angeles), and DemandGen (Montreal).

In the past few months, Lexalytics has been refining its sentiment engine. Catlin says that phrases like “it wasn’t the best movie ever” and “spectacularly bad” posed special challenges in terms of classifying their meaning. The company’s software has also gotten good at telling whether a statement is contextual rather than polar—the greeting “good morning” is the former, for example.

Of course, this is a very deep and long-standing problem that Lexalytics and others are tackling—getting a computer to quantify meaning and sentiment from phrases of text. Once the technology gets accurate enough—kind of like speech recognition over the years—we might expect to see some really big applications. One of them could be in mobile tech. Cell phone manufacturers are interested in things like summarization of e-mail, Catlin says, to make better use of small screens. You might imagine that software running on a phone (or in the cloud) could identify action items from an e-mail, and display those instead of the whole message.

As for Lexalytics, I wondered if getting acquired by a bigger player (perhaps one of its customers) might be in the works. Catlin says he certainly receives acquisition offers, but most of them aren’t serious. There’s “nothing active in that right now,” he says. On the other hand, he says, “I think the space will consolidate. Salesforce, Oracle, and Microsoft are buying things left and right.”

And he left the door open a little more. “We’re a little company,” he says, “and the eventual goal is find a home for the company in the long run.”

Gregory T. Huang is Xconomy's Deputy Editor, National IT Editor, and the Editor of Xconomy Boston. You can e-mail him at gthuang@xconomy.com or call him at 617-252-7323. Follow @gthuang

By posting a comment, you agree to our terms and conditions.