How do the employees in your company like the recent change in health benefits? Yammer can now offer insights into these questions via an integration with sentiment analysis provider Kanjoya.

Using Yammer’s open APIs, Kanjoya’s Crane analytics dashboard captures data from Yammer groups to take the emotional temperature.

80 Separate Emotions

In the case of an HR manager who wants to determine if the new health benefits are a hit or a bust, the dashboard can characterize expressive statements into 80 separate emotions, including “surprised” or “annoyed.” The dashboard can also present emotional assessment that is limited to a given team or location.

Kanjoya’s Crane for Yammer includes keyword search of analytics for a specific topic, and a Sentiment Graph can compare the volume of such emotions as joy, excitement, surprise or anger.

There’s a word cloud, so that specific words associated with trending conversations can be visualized as to relative importance. Also featured: an influencer analysis, which shows the most replied to, the most praised and the most liked individuals.

Armen Berjikly, chief executive officer and founder of Kanjoya, said in a statement that the Emotional Intelligence engine uses the Yammer data and “delivers detailed, actionable analytics, changing the way companies monitor internal sentiment from subjective gut-feel to data-driven.”

All the Tricks

The kind of Yammer data that is available for this kind of emotional analysis is about to increase. On Thursday, the company acknowledged reports that it is testing out a new feature -- instant messaging private chats between individual users.

The emotional analysis engine is based on Berjikly’s earlier Experience Project, a social networking environment where members write stories about who they are, how they feel and what they’re doing. On its Web site, the company said that the engine is “aware of all the tricks we use to express ourselves online,” because it is “sensitive to human emotion, not just a simplified notion of positive/negative.”

The engine returns a probability distribution across emotions, such as its characterization of the following written expression as 56% annoyed, 18% disappointed, and 11% confused, “ugh someone just spilled icee all over me at Six Flags.”