Overall, the CCC Council found the topic of geospatial computing to be an interesting and promising direction for catalyzing new work in computer science, and believes such advances could in turn propel this multidisciplinary realm of work forward.  The Council also found the high-level goals of spatial-temporal computation, transforming the Internet, and impact on science, as well as the sample focus problems mentioned (eco-routing and enhancing air-traffic control) interesting.

However, the CCC Council was concerned about whether the visioning exercise would focus on highlighting the promise of geospatial challenges to stimulate new thinking in computer science, or on applying existing methods in relatively straightforward ways to geospatial applications.  There seems to be some emphasis on the latter and less clarity and focus on the former -- and this was the weakest aspect of the proposal.  There was opportunity to include additional challenges at the intersection of geospatial concepts and computer science, and we think inclusion of a larger constellation of directions would help flesh out hard challenges in doing new computer science research to enable solutions.  Sample areas for promising work include links between geospatial computation and privacy and security, ecommerce, ideal sensing configurations, predictive modeling, and challenges in the geosciences.  In particular:

- Privacy and geographic information is an important topic:  while location information (per GPS on people’s phones, cars, etc.) can provide great value to users and industry, streams of such data also introduce concerning (and at best “spooky”) tracking.  Computer science efforts to date at obfuscating location information have largely met with negative results (see references below).  Addressing the challenge of leveraging useful geospatial information while protecting individuals’ privacy will likely promote and leverage advances in security, cryptography, databases, and HCI.

- The use of geospatial information in ecommerce is in its infancy.  Great opportunities lay ahead in the leveraging of users’ locations and expected routes in proactive services and assistance, ad impressions, and healthcare (again, see references below).

- Another challenge area is in the use of geospatial reasoning in sensing and inference across space and time.  Multiple tradeoffs (including those arising in privacy considerations) can come to the fore with attempts to sense and draw inferences from stable or mobile sensors – and the work extends to crowd-sourced sensors, such as the use of GPS sensors in portable devices for informing traffic work.

- Geospatial information can also be used to make predictions about everything from the next location of a car driver to the risk of forthcoming famine or cholera.  Such predictions would challenge the best of machine learning and reasoning algorithms, including directions with geospatial time series data.  We see rich problems in this realm.

- Eco-routing seems like a promising area for research and development as well.  We also see promise in generalizing eco-routing to include ridesharing and transportation planning challenges.

- Another direction more broadly is in supporting the constellation of problems in the geosciences – the core evolving basic and applied sciences of geographical and geospatial studies.  Challenges and characterization of computational needs in the geosciences can be found at http://www.nsf.gov/geo/earthcube/, and related resources are available through the NSF’s Directorate for Geosciences (http://nsf.gov/dir/index.jsp?org=GEO).

In addition to the research opportunities, the CCC Council had questions about the design of the workshop events and overall budget.  It is not clear that having three separate symposia constitutes the most productive way to succeed.  For example, it seems unnecessary to have separate meetings to deliberate about “needs” and “S&T Trends.”  There were also concerns about setting aside money to do outreach missions to other conferences.  Rather, a strong organizing committee may be expected to directly build an agenda for a “synthesis” meeting on key problems and promising opportunities – based on expert knowledge.

The CCC Council was generally impressed by the productive and thoughtful people included on the organizing committee.  However, we recommend that the group of organizers be bolstered with key government (e.g., NOAA, NASA, NRO, etc.) and industry participation (e.g., Garmin, Google, Microsoft, Lockheed, etc.).  This is especially important given that the role that government agencies and commercial entities play in fielding and relying upon geospatial applications.  Also, it would be good to add a strong person involved in the geosciences to the committee, and we recommend Amy Apon at Clemson University for this role (and would be happy to contact her if you agree).

We hope this feedback is helpful to you.  If you have questions, please contact Eric Horvitz (horvitz@microsoft.com) to discuss further.

Again, thank you for your proposal and for your interest in conducting a CCC visioning exercise in geospatial computing.  We look forward to seeing a revised version of your proposal shortly.

Sincerely,

Ed Lazowska, CCC Council Chair

Susan Graham, CCC Council Vice  Chair

Lance Fortnow, CCC Visioning Subcommittee Chair

Eric Horvitz, CCC Council member

Erwin Gianchandani, CCC Director

--
 
Several relevant references on additional topics:
 
Privacy and geographic information:
 
John Krumm, "Realistic Driving Trips For Location Privacy", Seventh International Conference on Pervasive Computing (Pervasive 2009), May 11-14, 2009, Nara, Japan. (PDF)
 
John Krumm, "A survey of computational location privacy", Personal and Ubiquitous Computing (2009) 13:6, pp. 391-399. (link) (PDF preprint)
 
John Krumm, "Inference Attacks on Location Tracks", Fifth International Conference on Pervasive Computing (Pervasive 2007), May 13-16, 2007, Toronto, Ontario, Canada. (PDF))
 
Sensing configuration and tradeoffs:
 
Andreas Krause, Carlos Guestrin, Anupam Gupta, Jon Kleinberg, "Robust Sensor Placements at Informative and Communication-Efficient Locations" [long version], In ACM Transactions on Sensor Networks, vol. 7, no. 4, 2011.   
 
Matthew Faulkner, Michael Olson, Rishi Chandy, Jonathan Krause, K. Mani Chandy, Andreas Krause, "The Next Big One: Detecting Earthquakes and other Rare Events from Community-based Sensors", In Proc. ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN), 2011.  [pdf]
 
Kause, E. Horvitz, A. Kansal, F. Zhao. Toward Community Sensing, Proceedings of IPSN 2008, International Conference on Information Processing in Sensor Networks, St. Louis, Missouri, April 2008.
 
Ecommerce based on routing and location:
 
E. Kamar, E. Horvitz, C. Meek. Mobile Opportunistic Commerce: Mechanisms, Architecture, and Application, Proceedings of AAMAS 2008, Estoril, Portugal, May 2008.
 
Ridesharing:
 
E. Kamar and E. Horvitz. Collaboration and Shared Plans in the Open World: Studies of Ridesharing, IJCAI 2009, Pasadena, CA, July 2009.
 
Predictive modeling:
 
J. Krumm and E Horvitz. Predestination: Inferring Destinations from Partial Trajectories, UbiComp 2006: Eighth International Conference on Ubiquitous Computing, September 2006.
 
A. Kapoor, N. Eagle, E. Horvitz. People, Quakes, and Communications: Inferences from Call Dynamics about a Seismic Event and its Influences on a Population. Proceedings of AAAI Symposium on Artificial Intelligence for Development, Stanford CA, March 2010.