When first learning the relational data model, I expressed frustration to my supervisor, complaining that some entity just wouldn't fit into the structure. He responded by saying, "anything can be stored in a database. Keep trying." While anything might be storable in a relational database, getting it out in a comprehensible fashion remains a problem.
The display of query results-especially large or fuzzy result sets-has been a problem that has interested me from the beginning. SQL in general has a problem with vague relationships-there is very little room for 'close'-either a record is in a result set or it is not. There is a field, however, that uses all the logic of the relational data model but fully accepts 'close': Geography. In the physical world, "close" is a perfectly knowable concept.
With this in mind, I approached Nicholas Chrisman from the Geography Department here at UW and discussed the possibility of exploring spatial data as a representation for textual data. He agreed to take me on for a one quarter independent study during which I would participate in an undergraduate cartography class (Geography 360: Introduction to Cartography) to learn the basics of cartography, the language used to describe the physical world, and the technology of Geographic Information Systems. Simultaneously, I was to engage a mapping project of some scale to investigate the technological and conceptual issues involved. The results are included in my paper, "The Central Eurasian Information Resource: Recommendations for a formal design process."
I was able to continue my studies with Dr. Chrisman in Geography 460: Geographic Information Systems Analysys, in the fall of 2001. This class allowed me to learn ARCview, the campus standard for GIS software, on a more advanced level, adding several more sophisticated methods of data analysis to those learned earlier. The major project from that class, Hazardous Waste Site Reclamation in the Bremerton Area, shows hazardous waste sites in Bremerton classified on an ordinal scale according to priority for remediation. While the statistical analysis itself is somewhat facetious, the process is a good example of visual analysis of data. This project was completed with two other graduate students--an urban planner and epidemiologist--who were electing in the geography department. One of the intermediate steps in the data analysis is shown, much reduced, here:
As part of my graduate assistantship this year, I was able to apply some of this knowledge to a real world problem. I was asked to design a survey instrument which could be used to measure the penetration of advanced telecommunications services on American Indian Reservations. It is widely acknowledged that both ethnic minorities and rural communities both lag behind the general population in the use of computers and the Internet. At the same time there has been no precise measurement taken to show exactly how far behind Native American communities are. Existing data is lacking for a number of reasons outlined in the documents referenced below.
I have taken to calling the requested metric 'native connectivity.' As part of the survey design, I have suggested that a centralized GIS repository be built to track connectivity projects and link them to general infrastructure projects on reservations. The accurate gathering of infrastructure information required for the survey would serve as the data foundation for this repository, and the tool could be used by reservation governments to identify continue to track projects in the future.
Another major argument used in my research proposal is that the information can not be gathered without laying a significant data-gathering infrastructure. The lack of this infrastructure is what has hindered previous attempts to measure native connectivity. Some preliminary maps used to illustrate the project are available.
GIS skills acquired: