Next time you check in with foursquare or Facebook, please stand right outside the venue, make sure your smartphone has a very good GPS signal and describe the location accurately.
That's the wish of the spatial data community, which is getting grumpy about user-generated spatial data.
To understand why, consider the several US restaurants called “Mom's Kitchen”. All are real establishments, unlike the many foursquare locations with the same name created by folks eating a home-cooked meal in their actual mother's real kitchen.
Jonathan Barouch, founder of augmented reality service Roamz, had to find a way to figure out which Mom's Kitchens were commercial operations and says this kind of labelling problem is common.
“Whenever you allow users to create locations and tag them you get issues,” he says. GPS accuracy causes some issues, and when combined with sloppy users exacerbates matters. “Someone checked into our office with the right address but from the other side of the street,” Barouch explains. The result is a search result nearly 100 metres from Roamz's actual office.
“We do not have administrator rights over Foursquare's data or we would go in and correct the address,” he says. “It is very difficult to fix because it looks like a valid address.”
Events present another challenge. Sydney's annual Gay and Lesbian Mardi Gras parade traverses around 500 metres of one street, but Barouch says many social media users checked in at a location they simply called “Mardi Gras.” Interpreting that location and the many different coordinates all associated with the same name is a tricky task for any provider of location-based services.
Other user-generated locations are just junk.
Jennifer Wilson of digital production company The Project Factory points out a Sydney location named “I'm on a horse” as an example of junk data.
Deliberate false naming is another issue, as Wilson says foursquare is being gamed by users desperate to acquire badges. Wilson tells of locations where users create alternate names for a venue so they can be made its Mayor. At other locations, several people inadvertently give the same place a different name. Sites like foursquare make no attempt to curate that data, she says, because they want to cash in on the overall volume of check-ins.
This makes life difficult for The Project Factory when it builds locative games or other location-based services, as it becomes hard to understand players' true location. The company has found a fix that compares a user's professed location to the actual coordinates of an immovable nearby landmark.
Roamz has also developed a set of workarounds that check multiple public sources of spatial information to boost the accuracy of its services. “We look at Twitter, Instagram, foursquare and map the name of the location. If we have a longitude and a latitude we reverse geocode it. It's a challenge and we have not got it right yet.”