UCGIS Virtual Seminar - Fall 1998 [Back][Refresh][Options][Search] Part 3: UCGIS Approach [Edit*][Delete*] [Image] Part 3: UCGIS Approach - Discussion Dawn Wright 09/27/98 Topics [Image] Object-oriented Design Erik Shepard 10/07/98 [Image] Multidimensional Representational E. Lynn Usery 10/07/98 Theory [Image] [Image] re: Multidimensional Ronald William Ward 10/09/98 Representational Theory [Image] [Image] Re: representations and Erik Shepard 10/09/98 dimensions [Image] [Image] Beyond Spatial Fetishism and Byong-Woon Jun 10/09/98 Toward MRT [Image] Post new message in this thread ------------------------------------------------------------------------ [Top][Previous][Next][Print][Reply][Edit*][Move*][Delete*] Date: September 27, 1998 09:45 PM Author: Dawn Wright (dawn@dusk.geo.orst.edu) Subject: Part 3: UCGIS Approach - Discussion Topics UCGIS Approach - Discussion Topics will include: Multiple representations --Various geometries for location-based representation --Combination of location-based, feature-based, temporal-based Tools - How do these help --Computational geometry --Object-oriented design Representation for data mining and knowledge discovery Conceptual development with empirical testing Broader theory of geographic representation New multidimensional representational theory reflecting human cognition and efficiency for computer implementation (http://forums.library.orst.edu/forums/Index.cfm?CFApp=7&Message_ID=1569) ------------------------------------------------------------------------ [Top][Previous][Next][Print][Reply][Edit*][Move*][Delete*] Date: October 07, 1998 08:42 AM Author: Erik Shepard (shepard@uga.edu) Subject: Object-oriented Design To address one of the topics in this section, my thesis is (among other things) looking at using a single geodata class to represent all geographic data. While this does not help with how to store the data at the most fundamental level, I believe that it does conceptually simplify the confusion proliferating now. Basically, the geodata class encapsulates vectors, rasters, and whatever else we may come up with now or later. To the user, they interact with the data the same way (that is, via the predefined methods) regardless of the representation. This simplifies the way in which the look at things and they don't have to worry about the format that the data is in. (http://forums.library.orst.edu/forums/Index.cfm?CFApp=7&Message_ID=1765) ------------------------------------------------------------------------ [Top][Previous][Next][Print][Reply][Edit*][Move*][Delete*] Date: October 07, 1998 12:07 PM Author: E. Lynn Usery (usery@uga.edu) Subject: Multidimensional Representational Theory One of the long term UCGIS objectives of this research topic is the development of a multidimensional representational theory for geographic phonemona and processes which captures human cognition and is also efficient for computer implementation. As we approach the end of our two-week discussion of Geographic Representation, I would appreciate hearing from seminar participants on this topic. Is such a theory possible (consider that geographers have been trying to develop "theories" of spatial location and distribution for many years without tangible success)? How can such a theory incorporate the seemingly random aspects of human action? Is is sufficient to simply develop a theory for geographic phenomena or must human action and processes be included? Some initial work towards such a unifying theory can be found in references by McMaster, Nyerges, Peuquet, and Usery in our reference list. (http://forums.library.orst.edu/forums/Index.cfm?CFApp=7&Message_ID=1775) ----------------------------------------------------------------------- [Top][Previous][Next][Print][Reply][Edit*][Move*][Delete*] Date: October 09, 1998 08:23 AM Author: Ronald William Ward (ronward@arches.uga.edu) Subject: re: Multidimensional Representational Theory Can a multidimensional representational theory (MRT) capture the seemingly random aspects of human action? In the past, and in two dimensions, geographers like Christaller (central place theory) and Von Thünen (model of agricultural land use) have attempted to capture random human actions and generalize these actions into informative representations. It seems to me that with the multidimensional approach we are trying to generalize less, and instead, work towards more accurate representations which include explicit third and fourth (time) dimensions. Is there a contradiction here? When a set of phenomena like human action is seemingly random in nature, it would appear that the only way we can make sense of the random dynamic invloved is to generalize patterns from it. Yet, the multidimensional approach, in that the goal is to develop more accurate representations showing the true complexity of space-time dynamics, might be moving away from a theory or set of theories that will aloow us to make sense of random human action. Where is the compromise? Part of the difficulty may lie in the extent to which our thinking is unfluenced by the neoclassical econimic paradigm. Since Adam Smith and The Wealth of Nations we have been subject to a system that assumes everyone makes decisions based on rational economic behavior, and that everyone will try to maximize benefits and minimize costs. How does a pure science directive such as formulating a MRT operate within the constraints of funding opportunities and assumptions that may not apply to all aspects of random human action? In a world where policy decisions (remembering that main reason for development of GIS was and remains as an instrument for making policy decisions) are projected for short-term maximization, and long-term ideas often disolve in political limbo - will space-time theory be generally applicable for use in formulating policy? It's not a matter of whether or not we can formulate such a theory, it's a matter of whether or not the goals of the GIS community at large are ready to commit to formulating such a theory. Human action and processes must not only be included in the formulation of a MRT, but must also be overcome so that formulation of a MRT will be possible. Ron Ward (http://forums.library.orst.edu/forums/Index.cfm?CFApp=7&Message_ID=1831) ----------------------------------------------------------------------- [Top][Previous][Next][Print][Reply][Edit*][Move*][Delete*] Date: October 09, 1998 09:06 AM Author: Erik Shepard (shepard@uga.edu) Subject: Re: representations and dimensions One of the problems that I see is that we not only need to address multidimensional representations, but also multirepresentational dimensions (pardone the word-play). We have in GIS right now the raging raster-vector debate. Any extensions to these representations seem to me to be further widening the gap. For instance, to consider spatial, 3 dimensional data, in rasters we could implement voxels and in vectors we could implement something akin to a 3D shape (i.e. {x, y, z} coordinates). Adding time, we must extend even more by perhaps adding {x, y, z, t} or something worse and creating perhaps 4dimensional voxels. Or perhaps we must consider to mathematical modeling approach. In any case, we need to look towards unifying the data structures before we begin to extend them. How do we do this? I have no idea. But I think that it is something that needs to be considered. We have been trapped for too long in Berry's geographic matrix. I don't know what sort of paradigm shift we would need to embrace to break out of this, but it seems like all of the work to add time (Langran, etc) and 3 dimensional data models (Pequet) is limited by the constraint to this matrix. Just some food for thought. (http://forums.library.orst.edu/forums/Index.cfm?CFApp=7&Message_ID=1834) ----------------------------------------------------------------------- [Top][Previous][Next][Print][Reply][Edit*][Move*][Delete*] Date: October 09, 1998 02:02 PM Author: Byong-Woon Jun (bwjun@arches.uga.edu) Subject: Beyond Spatial Fetishism and Toward MRT From the viewpoint of critics to geography, it seems to be true that there have by far been several efforts to develop geographical theories without visible success. However, I think we, geographers, have had two major theories before and after 1950's: Central Place Theory by Christaller and Spatial Diffusion Theory and Time Geography by Hagerstrand). The former is based on static and deterministic approaches while the latter dyanmic and probablistic approaches. It is generally accepted that these two theories are the best-known efforts within the field of geography. Especially, diffusion theory and time geography tried to incorporate the seemingly random aspects of human action such as the spread of AIDS and daily activity patterns through generalization and abstract sequence of the reality. Diffusion theory focuses on the overall pattern of specific natural or cultural phenomena as change spreads through space over the passage of time. Time geography deals with space-time patterns to the individual level observing paths of individuals through space and time and their interactions. However, such a theory faild to fully represent the complex human action which is random and dynamic in nature because of limitations of traditional cartographic representation. It is urgent that we need not only to continue examination of social process such as daily activity patterns and uraban growth, but also to advance understanding of the effects of human activities on the natural environemt. In this regard, it is necessary that we must include human action and processes into geographic representation theory and that we need to develop a new multidimensional reprepresentation theory reflecting such human cognition and efficiency for computer implemenation. (http://forums.library.orst.edu/forums/Index.cfm?CFApp=7&Message_ID=1838)