Here are some notes from the Oracle location intelligence conference in Washington, DC.
People here are talking about fun things one can do with location data. It mostly boils down to a) where you have been, b) where you are and c) where you will be, and then using that information to make better decisions. There are case studies from the tech team that helped reelect president Obama, to those who analyzed social media to help you find a better job nearby. Some case studies show how these tools can be used to advance the public good, others just how to improve marketing ROI. There are various buzz words buzzing around, #geosocial, #geovalidation, #geomarketing, #geotagging, #geoanalytics,.. There are many companies, big and small, generating new business by figuring out new ways on how to do what we do, a little bit better.
One thing they all have in common is “geodata” which is sometimes also known as “big data.” Without the availability of geodata, none of these companies would have gotten off the ground in the first place. They are mostly US-based companies and they export their know-how and technology all over the world.
I was talking to a representative from a company offering “cloud geocoding services.” They had some interesting views on their coverage for Canada which is only “partial.” The main reason it seems, is that licensing of geodata in Canada is much more complex than in other countries (The US according to him is the best). In the view of my colleague “Canada is the most difficult developed country when it comes to obtaining & licensing geodata.” The same view was shared by several others.
Canada’s location data are hard to get for small companies, but big companies still get what they need since they can pay the asking price. There is Oracle’s geocoding suite, Google Maps, Microsoft’s Mapinfo, ESRI and others that provide you with all the geocoding you need for Canada.
However, a global geocoding engine can not fully replace a customized local solution. Take google’s geocoder for instance. Google approaches the problem of geocoding like a search problem. They return the highest ranking location from their database. It could work great in certain countries but not all. Because different countries write location information differently. A brazilian colleague pointed out that they had to build their own geocoding engine, which is much more accurate than any of the global geocoders mentioned above.
Case in point. Can Google geocoder accurately find my parent’s house in Ottawa? As of this moment, no. Geocoder.ca can(1). According to one presenter only 44% of companies are satisfied with results of the geocoding process (of those companies using the well established players), 37% said it is too unreliable, 27% takes too many resources, 14% it is too slow.
This is not just an attempt at self promotion. It is just to state the fact that with even imperfect data one can build a geocoder that is of comparable accuracy to a geocoder utilizing the best geodata money can buy. If you focus on a small localized problem, chances are you might find a better solution than someone trying to solve all such problems with a single algorithm.
Location data are facts, but it costs money to collect those facts. Various companies collect this data in the course of their activities at a great expense (Google, UPS, FedEx, etc). The government spends a lot of taxpayer money for the same goal. But the government can never make that money back by selling the data at high prices. It is better to just give it away for free. The US government does this. It generates more innovation and more companies and more business, which in turn brings in more tax revenues.
Government, is the only big data company in the position to release that data to the public at no cost, then sit back and earn revenue as a result.
I hope our Canadian government realizes this someday.