At the COM.Geo conference today held at George Washington University in downtown Washington DC, there were several presentations on the "Internet of places and things." While none of the information presented by the speakers was new or earth-shattering, the body of material was more about the patterns of location data and a more semantic approach to search to understand those patterns.
In a presentation by Giuseppe Conti of the Fondazione Graphitech (Center for Advanced Computer Graphics Technologies) in Italy he wanted to better enable the "Internet of Places" with a virtual structure of space-time-tasks to find and use Internet resources more easily. He believed that when crawling the web for information that searches should be implicitly spatial. In fact, later in the day, Richard Barnes from BBN pointed to the obvious case where if you used a browser in a public place such as a library, and searched on "Starbucks," for example, the results would automatically suggest those nearest to the location where the person issued the search. This kind of search is not brokering location from a device like a cell phone but rather something like the browser's IP address or perhaps even other searches from that browser.
Conti also suggested that searches on the Internet should be indexed to accommodate the "vast range of scales" defining "different spatio-temporal tolerances." The implication for both of these examples is that today's Internet searches are entirely inadequate . Both Barnes and Conti where suggesting that Web search architecture needs to change and Barnes suggested that location services based on context is much more relevant.
Kipp Jones, chief architect at Skyhook Wireless, presented his thesis on how data from aggregate user behavior could be used. Skyhook Wireless collects historical and real-time data from the aggregate uses of its location server. These data can be viewed as thematicized surface or "heat" maps. Baselines of these patterns could be cataloged so that anomalies to these patterns might show some unique demographic characteristic. For example, what might these data indicate about "groups" of users (think "flash mobs") patterns associated with "events" like a sports event. Skyhook has marketed these "SpotRank" to retailers and others in hopes of getting them interested in looking at a new kind of demographic data that can be gleaned from aggregate, anonymous location requests. Could retailers react to this new kind of search that reveals data for which they could offer coupons and other offers.
Finally, in a presentation on another kind of data search, Jon Gosier of Ushahidi presented information about using QR codes in crisis events. To Gosier, QR codes are very much like "check-ins." They would be easy to integrate into a well-coordinated crowd-sourcing project and a quick and effective way to augment participant experience. But Gosier suggested that QR codes could be used in a few different applications:
- Sending instructions or transmitting messages
- Triggering a remote process (API call)
- Triggering a process on the device itself - like triggering a phone to send a message to people who do not have smartphones
- Authentication; the QR code can serve as an identification measure
- Logistics/resource allotment - sender may know what resources are available at sender location
So, I took away from this first day that more people are thinking about the problem of location-based context searching and the mechanism by which we explicitly or implicitly use location in search criteria.