AT&T mines Twitter for anger over dropped calls

Nothing infuriates a cellphone user more than the rude, abrupt experience of losing a call, and no device has attracted as much ire for its call-dropping tendencies as the AT&T-carried iPhone. And generally speaking, doesn't it seem a bit absurd that our flashy new future-phones are worse at making calls than the rickety old landline phones that we all grew up with? Anyway, sorry, enough with the complaining...

Or maybe not. AT&T is experimenting with a technology that actually parses the angry Tweets of thousands of its users for clues.

So, how does AT&T convert raw material like this:

Into useful data? With a little software magic. Technology Review reports:

[AT&T] uses two levels of filtering to find tweets by frustrated customers, and they do this by tapping into the programming interface tools Twitter makes freely available. A general set of queries pulls in every tweet related to AT&T's mobile service before a more rigorous set of rules homes in on those relating to service quality, for example messages containing words like "call dropped" or "3G." This automated method was around 90 percent accurate at identifying genuine complaints, the researchers found.

Once the Tweets are identified, AT&T can use attached location data and timestamps to zero in on the circumstances surrounding a dropped call, or at least approximate its time and location.

AT&T has a fairly extensive internal monitoring system, which returns data about call failures and successes, but such information lacks a certain... human element. A small spike in dropped calls in a particular area in Las Vegas, for example, doesn't tell much of a story. A string of Tweets from a local trade convention lamenting a near total lack of service, however, does. Mining this data is less about building a database of raw network performance data than it is about injecting the customers' voices into it.

This isn't AT&T's first foray into crowdsourced network reporting. In December of 2009, the company release an iPhone app (and later an iPad app) called Mark the Spot, which let its customers file instant reports of a dropped call--assuming their connection is strong enough to submit them.

Reports from an app, or on Twitter, are fundamentally different from those monitored inside the company. A call that couldn't even be placed might not register with AT&T's hardware, but could result is an exasperated Tweet. Vague complaints about voice quality or data reliability might be hard to quantify or track, but are exceedingly easy to put into words. ("This sucks!")

Obvious jokes aside, AT&T has used the data harvested by Mark the Spot to improve its coverage. (Or, at the very least, to communicate to affected users about preplanned upgrades.) Data gleaned from Twitter won't yet be used in an operational sense, AT&T says, but the company hopes to refine its methods to the point that Twitter data can be combined with Mark the Spot reports and internal tracking to paint a much more complete picture of where the network works, and where it doesn't.

This post was originally published on Smartplanet.com