Microsoft's 'MoodScope' phone software senses your mood
Your phone is there, always watching. It’s your life companion, according to one manufacturer. So shouldn’t it be able to tell your mood, too? Microsoft thinks so.
In a research paper, Microsoft said that by analyzing phone calls, texts, the browser history, and other common smartphone interactions, a new MoodScope service it developed could accurately predict the user’s mood 93 percent of the time after a two-month period. This was after the phone was “trained” to sense the user’s mood.
So what good is it? After all, you don’t need your phone to tell you how you feel. Instead, the idea is that your phone’s new “mood sensor” will tell others how you feel—social networks, your friends, even your mom.
Furthermore, that information could be passed along to services like Spotify, which could curate an emo-weighted playlist for when you’re down in the dumps. Microsoft even created a “MoodScope social-sharing application” to share users moods to their Facebook Timelines.
Yes, this is for oversharers. Microsoft isn’t proposing that all phones should sense your mood, automatically, but that it would be based with an app and tapped into via an API.
Somewhat ironically, researchers have never tested the technology with a Windows Phone; they used a combination of Android phones and iPhones for the study, using 32 participants from the United States and China.
”We foresee mood inference as a vital next step for application context-awareness,” wrote Robert LiKamWa, Yunxin Liu, Nicholas D. Lane,and Lin Zhong, the co-authors of the study. All, except for Rice University’s Zhong, worked for Microsoft Research. “Such inference would improve the utility of the smartphone and lower the social barrier for sharing mood.”
How does it work?
Specifically, the researchers used SMS, email, phone call, application usage, web browsing, and location to determine mood, defined as a persistent emotional state, rather than flashes of one emotion or another. Microsoft naturally used both the phone and a cloud service to collaboratively produce its MoodScope responses.
What the researchers found is that determining mood wasn’t easy; they first had to ask the users to record what their moods were every four hours, and use that to determine a general mood model.
Over time, however, the general mood model cut down the training time to about ten days, during which moods were sensed with 72 percent accuracy.
The paper does not draw any specific conclusions about how smartphone use is tied to mood, such as happy users frequently accessing their phones. But the paper does conclude that phone calls, certain categories of applications, and locations are often tied to a pleasurable mood.
As academic papers often do, the authors are conservative in their conclusions, noting avenues for further research—such as, for example, rises in frustration levels tied to heavy traffic.
MoodScope sound innocent enough. But also keep in mind that gauging emotional response is a key component of assessing the effectiveness of advertising. In 2012, Microsoft filed for this patent on using sensors to assess emotion—imagine that a Web page that knew you were sad might target you for an ad pushing comfort food, for example.
Yes, mood and emotion are two different things, as the paper points out. But there’s always a tradeoff; on one hand, knowing your mood may provide better recommendations for music and other services. But they can also be used against you.
For comprehensive coverage of the Android ecosystem, visit Greenbot.com.