Hyper-Local, Geocoded Sites Need Community Input

New York, N.Y.
If you've tried out any of the various geocoded, hyper-local media services, you've probably noticed that their data is still fragmentary and not always accurate. The problem is not with the geocoding technology itself, which as long as it is within range of the satellites that enable it to establish latitude and longitude always functions just fine.

Rather, the problem is with the government records, maps, telephone listings, and other grounded databases, which are often months, years, or even decades out of date, as well as with misspellings and other human errors on the ground.

Everyblock founder Adrian Holovaty described one such glitch recently on his company blog: The Los Angeles Police Department's crime maps contain address inaccuracies that, in turn, had been replicated on Everyblock.com, so Holovaty and team implemented a new safeguard to try and adress this problem.

Everyblock is now comparing the LAPD's geocoding for crime locations against its own geocoding of the city's grid. If there is a substantial difference, "we won't geocode that crime at all, and we publish a note on the crime page why a map isn't available."

Holovaty went on to explain that his service is extending this methodology to all of its public data sources, and that they are adding a category, Unknown, to capture all of the records that don't match up correctly, as well as all data that remain un-geocoded. To get a sense of the scope of these data gaps, he said that crime maps in Philadelphia over a one month period indicated that about one percent of crimes occurred in unknown locations.

Everyblock is hardly the only venture struggling with these types of data gap issues. A number of the photo and blog aggregation services I've tested have also failed to accurately pinpoint the locations of photographers and bloggers on their maps. When I've contacted customer service at these companies, seeking an explanation, I've usually received one of those "We're working on it" messages.

Everyblock, which is funded by a $1.1 million grant from the Knight Foundation, has created sites for 11 U.S. cities, including New York, Seattle, Chicago and San Francisco, and is one of the best hyper-local media extensions out there so far. Its decision to double-check source data is a great first step in the battle to provide more relaible information to users..

The next needed step, I believe, is to invite the community to help out. As we've seen in many self-regulating spaces online, the community collectively knows much more than the mere sum of its parts.

Besides, who doesn't like to help solve a crime? Or, at least, help identify where it took place?

---

Note. This item was posted from the following location: