Twitter can't do much to keep your privacy when the content of your tweets gives you away

Mar 24, 2014 15:18 GMT  ·  By

Whether you allow Twitter to attach a location to your messages or not may not be that relevant for marketers, especially if they have other ways to figure out where you’re sending messages from.

ArsTechnica reports that three researchers from IBM have managed to develop an algorithm that can be used to predict a Twitter user’s location without actually needing a geotag.

This is done by comparing the content of an individual’s tags to those made by other people who do have enabled the geotag option. This turns out to work in most cases, making the algorithm a pretty good prediction tool.

Aside from the geotags, tweets can also contain various hashtags, as well as text references to cities, landmarks or states, to name just a few, something that can help locate anyone quite easily.

Researchers analyzed about 1.5 million tweets, out of which 90 percent were used to train the algorithm, while the rest of 10 percent were used to test it out.

Researchers could predict which city someone was in with a 58 percent accuracy, while larger regions were predicted with better accuracy, namely 66 percent on a state level and 73 percent on a time zone level.

“First, the output can be used to create location-based visualizations and applications on top of Twitter. For example, a journalist tracking an event on Twitter may want to know which tweets are coming from users who are likely to be in a location of that event, vs. tweets coming from users who are likely to be far away. As another example, a retailer or a consumer products vendor may track trending opinions about their products and services and analyze differences across geographies. Second, our examination of the discriminative features used by our algorithms suggests strategies for users to employ if they wish to micro-blog publicly but not inadvertently reveal their location,” the researchers explained in their paper.

Tweet locations are particularly important to marketers who can direct messages at people found in proximity of their shops. The bad news for those who want to maintain their privacy is that they can do very little to avoid others getting a general sense of their location.

While sending messages without attaching any type of geotags is a great way to wrap yourself in a veil of privacy, there’s nothing you can do about an algorithm such as the one developed by the IBM researchers who judge the content of your messages rather than the metadata.