Canvs is a startup measuring emotion on social media, but CEO Jared Feldman insisted that we shouldn’t think of it as a social sentiment company.
He said that’s because existing tools for measuring sentiment have earned something of a bad reputation. Where those tools are usually limited to classifying social media posts as good, bad or neutral, Canvs can classify a statement into 56 different emotional categories, including like love, dislike, annoying, boring or crazy.
Canvs announced today that it has raised $5.6 million in Series A funding led by KEC Ventures. Other investors in the Series A include Rubicon Venture Capital, Gary Vaynerchuk and BRaVe Ventures, Social Starts and Milestone Venture Partners.
The company’s big goal is to update the way television networks, media agencies and other businesses measure audience response to their content — not how many people watched, but rather how they felt about it.
So instead of calling a handful of people into a room for a focus group, you get a timeline showing the breakdown of all the different emotions expressed on social media around given piece of content. You can even connect that data directly to the moment in the show or the ad that provoked a certain response.
Canvs customers include Sony Pictures, SMG, NBCU and Viacom.
Feldman said the company has built semantic and natural language technology that helps it parse the vagaries of how we all write on social media, with slang, misspellings, emojis and more. For example, he said it can understand phrases that sentiment analysis tools would struggle with, like “I can’t freaking stand how much I love this show.”
To do that, Canvs has built a data team led by Chief Scientist Sam Hui (also an associate professor at the University of Houston’s Bauer College of Business), plus an editorial team that deals with the “nuances” of language: “They spend a lot of time on Urban Dictionary, they spend a lot of time talking to people.”
Ultimately, Feldman said Canvs’ real strength doesn’t lie in any one platform or source of data, but in the broader problem of analyzing text for emotion.
“Social is our medium,” he said. “We’re using this as a means to understand how people feel about things, and that’s a much bigger problem.”