A brief summary of a workshop on Volunteered Geographic Information
Dec 13-14, 2007
Upham Hotel, Santa Barbara, CA
Check out the 44 participants and the contributed issues papers.
The presentations covered everything ranging from smart dust to vitual globes. Following are a few presentation summaries:
Sarah Williams from the Spatial Information Design Laboratory, Columbia University, talked about smart sensors for solving global problems such as cell phones that transmit geocoded ambient information, digital traces that we leave everywhere we go such a while swiping a subway card, crossing a traffic light, working at a wifi hotspot, or talking on the cellphone.
Rajan Gupta, Los Alamos National Labs, gave a presentation on GPS units that can be extended with low-cost measurement devices: for example, GPS that not only records water locations, but also measures water quality.
Sarah Elwood, University of Washington; David Tulloch. Rutgers University, talked about VGI from the grassroots where citizens contribute and fill in the gaps that the government can't or won't.
A case-study of the National Map was presented by Morgan Bearden, The National Map, USGS.
Steve Coast gave a presentation on OpenStreetMap as a specific case of organized VGI.
Volunteered personalized driving routes were presented by John Krumm of Microsoft.
Mike Goodchild, Spatial@UCSB, described Waldseemüller map as one of the first documented examples of VGI. In today's world, while a formal naming process for placenames exists, technology makes it possible to have multiple names for a single location. VGI itself is described by many different terms: user-generated content/collective intelligence/crowdsourcing/asserted information. Whatever it is called, it leads to empowerment of millions who are untrained and have no authority otherwise. VGI leads to non-uniform coverage as only “interesting” places tend to get covered, and depends on web search engines to allow us to find it. There are three types of sensors: inert or fixed; carried on moving objects; and human beings. A key trait of VGI is that humans act as sensors. This is really “citizen science” in action, and some of its examples are the Christmas bird count and Project GLOBE. Some possible research questions to consider are: Why do people do this? Is it self-promotion (exhibitionism, retaining “ownership” of contributed data); altruism; a desire to fill gaps in the available data; or sharing with friends? Studying the range of authority and assertion, the potential for subversion of information, and the review process which may or may not be localized.
Werner Kuhn, University of Muenster, gave a presentation that claimed that 80% of all decisions are based on spatial information. Like in any decision-making, information loops exist in geographic information based decision-making as well. Characterizing VGI quality: completeness, consistency. Notions of place, discovering VGI, integrating VGI and GI, grounding semantics, modeling trust and reputation, liability. Metaphors for web interaction, incentives, social semiosis with VGI. Scaling the loops: from geeks to everybody, from GPS tracks and images to rich data and services, from disconnected loops to interfaced loops, from a few big social networks to many small ones.
Jack Dangermond, ESRI stressed that there is room for both VGI and authoritative GI, for different purposes as well as to validate the former against the latter. One way to think about it is that VGI is “action driven” while GI is “process driven.” VGI is basically observational assertions and metadata about such assertions are very important.
I offered Amazon's “Real Name” feature as an example of metadata about assertors. David Maguire, ESRI, demonstrated their distributed GIS platform that allows loosely coupled authors and users, mashups, and use of standard APIs with ArcGIS as a system for authoring, serving and using VGI/AGI. ArcGIS server has a crawl-able, KML-tagged “Services Explorer”.
Jack summarized with his observations on the entire workshop. He commented on GIS and VGI relationship — how can GIS users use VGI data? How does GIS support VGI? Does VGI have the promise of SDI? How can we mine VGI data for experts use? VGI benefits greatly from GIS concepts — spatial referencing system, visualization and query tools, web servers and services, shared data bases. What would GIS professionals say about VGI? Well, a good basemap is important, data models are important, standard workflows to create, maintain, edit and manage data are important, good geographic data requires a lot of work, spatial analysis modeling requires consistent data models, VGI observation data and assertions are valuable but how do we organize and integrate? (Spatial data mining, ETL) Six types of geographic knowledge: geog data, data models, geoprocessing models, geospatial workflows, metadata, maps and visualization. Distinction between amateur and professional systems: LA street lights, NESA street lights (Denmark, allows neighbors to dim their street lights), DHS security, NYC 311, BLM surveys, WWF Forest Watch.
Lior Ron from Google, asserted that we are sitting on the long tail of geographic data (breadth: how many places we know; depth: how much do we know about each of those places). Google has counted seven million “My Maps” instances, 300 million Google Earth activations, more than 50,000 API sites, and estimates 1000 human lifetimes spent looking at satellite photos. They call this the emergence of a geoweb, and are working on creating a new geoweb search.
Per Allen Carroll, National Geographic, they are geoenabling their content. They demonstrated Meta Lens, a web based platform for managing geo-enabled content and talked about LandScope America (to be launched in 11 months) in partnership with NatureServe. NG believes that while we are in great shape as far as imagery is concerned, the GIS data are spotty albeit very rich. It needs to be better supported and aggregated. While GIS data are in a pretty good shape at small scales and getting better at very large scales, VGI might help fill in the “gap in the map” in between small and very large scales.
Ben Lewis, Harvard Center for Geographic Analysis, told that his group is embarking on creating an “Africa Map,” a one-stop shopping for Africa continent base maps, online atlas and index, a gateway to more specific data searches across multiple systems, search non-spatial visual data, and a repository for Africa research projects. There is a lot of data on Africa, but not many know about it. Africa has been mapped by colonial powers for over a hundred years. Most of the continent is LandSat (not very good imagery). Russians have the best mapping of Africa.
Don Cooke, TeleAtlas, observed that users of geographic information are two orders of magnitude greater after Google Earth than before.
I gave a presentation on the [Science Commons Data Mark](). At that time this was still an “upcoming data mark” which became official three days later! I have been involved in this initiative since the October 2006 workshop at the National Academies on “Information Commons of Science” followed by the May 2007 Brazil workshop on Open and Permanent Access to Scientific Information and the final workshop titled [Common Use Licensing of Scientific Data] in Paris in September 2007. The Paris workshop was really where most of the ideas of the Data Mark were crystallized, so I have been able to develop a presentation on [Licensing Scientific Data] that I am going around giving wherever I can.
Some of the many, many research questions that came out of the workshop:
The findings of the workshop will be published. The final outlet is not determined, but it might be a special issue of a suitable journal such as the International Journal of Spatial Data Infrastructures Research (IJSDIR) or the GeoJournal.