Say What? MIT Spin-Out Helps Businesses Translate Online Chatter.

I don’t want to startle anyone, but it turns out some people aren’t shy about sharing opinions online.

They’ll take to Twitter, chat boards and comment sections to offer brands, celebrities, politicians and even well-meaning tech bloggers their articulate, reasoned and constructive feedback. Or something like that.

But the sheer volume of online chatter makes it difficult for businesses or researchers to extract meaning from it, to know whether consumers are on the whole “sad,” “happy” or “other” that you, say, performed a surreptitious psychology experiment on hundreds of thousands of users.

Luminoso, a Cambridge, Mass. startup that grew out of MIT’s Media Lab, has developed artificial intelligence software that promises to help businesses understand and draw insights from these virtual conversations.

Luminoso will announce Tuesday that it has raised $6.5 million in a funding round led by Acadia Woods. It comes on top of a $1.5 million seed round from angel investors two years ago. Digital Garage also participated in the most recent round.

The company says it can cull through millions of communications collected from social networks and websites by its customers, identifying shifting sentiment, picking up subtext and translating slang. And the text analysis technology works in English, Portuguese, Spanish, French, Italian, German, Chinese and Japanese.

Early customers include General Mills, MARS, Intel, REI, Scotts, Conagra and Sony, which worked with Luminoso to analyze social media conversations concerning the World Cup.

The company plans to use the funds to expand its staff and accelerate growth.

“We’re encouraged by our investors’ support and are dedicated to helping our customers continue to gain insights from their data, so that they can grow and improve their businesses,” said Catherine Havasi, chief executive and co-founder of Luminoso, in a statement.

Update: This story has been updated to correct the amount of the earlier seed round, due to incorrect information initially provided by the company.

This article originally appeared on Recode.net.