Posts Tagged ‘ontology’

A Tangent from Fuzzy Footprints…

Thursday, March 15th, 2012

Goodchild’s (somewhat uncoordinated) introduction to Fuzzy Footprints got me thinking, once again, back to ontologies–as has been mentioned by many others posting not only on this topic, but on many topics we have covered in class this semester.  So it brought me back to another question asked in class, again with regards to multiple topics: how important is geo-education?  And so here I would argue: VERY important.

Uncertainty can be largely down to our ability (or lack thereof) to communicate, and to understand what has been communicated by others.  Boundaries, locations, and our ability to define them are essential to geolibraries.  If we cannot come to general understandings, there will constantly be error.  Before in class I was not convinced that education (about scale, particularly, but about various geographic phenomena) should be made explicit (outside of a geography class).  Now, I believe otherwise–how could I not after repeating topic after topic that ontologies (and thus understanding) is important?

To create a global database of georeferenced information is a magnificent endeavour.  To create a global database of georeferenced information that can be efficiently searched by any member of the global community is a whole new ballgame, and must necessarily involve a renewed goal of educating the public and of coming to shared understandings (both on areas of agreement and disagreement).

-sah

Gazetteer Issues and Interoperability

Thursday, March 15th, 2012

I’ve been thinking about “cyberinfrastructure” and “madskiier’s” posts discussing the difficulties of incorporating an appropriate “gazetteer” function with geolibraries and how this function might need to change rapidly online. However, I think we are also facing a much greater challenge that harkens back to our lecture on ontologies. A huge question that I have about this idea of a geolibrary stems from the various definitions different cultures might have for differing types of geographic features such as mountains. How we define such features and their boundaries will be an essential question going forward for any researchers looking into how to create a comprehensive geolibrary that can cross cultural, political and physical boundaries.

Michael F. Goodchild hints at this in Chapter 5 when he poses a number of possible research questions. In particular, I’m interested in his questions about how much and what kinds of metadata are needed to support a geolibrary (Question 5, Page 8 ) and what are the cognitive problems associated with using geolibraries (Question 7, Page 8). One of the keys to making a geolibrary useful and operable across the boundaries I mention above may be figuring out how to set standards for the data, and in supplying lots of useful information about the data. Metadata could serve this purpose, but, as Goodchild notes, the question may more be one of what system do we use to organize this data so that it maintains it’s usefulness and interoperability. Here, of course, you get deeper down the rabbit hole and have to begin thinking about who is going to host the geolibrary, what kind of infrastructure it requires, and, then, what kinds of systems it can support and where the data will come from.

This seems like a much broader question than just how do we search by place names or incorporate functionalities and parameters into such searches that can help diverse sets of users. Rather, I think we are beginning to ask fundamental questions of how do we define different parts of a geolibraries platform, it’s data types and the ways in which we interact with these data types. Questions of ontology and organization.  Interestingly, there is PhD student at Xavier University who has been thinking about these questions in terms of digital libraries too – and he writes that “[Digital libraries] also pose technical and organizational interoperability challenges that must be resolved.” Find more from him here.

–ClimateNYC

Practicality in, reality out? Sort of

Friday, February 3rd, 2012

Kuhn’s style in addressing ontologies differed from that of Smith and Mark’s. His article is more comprehensible, as it has more focus and attempts to cover less ground. However, I did find the articles to successfully complement one another. The main scope of Kuhn’s article, focuses on “problem-solving world knowledge” (with an emphasis on operations and domain theories), rather than “problem solving methods or reasoning”, is a step in the right direction (616). If ontologies will be diversified, inquiring about knowledge similarities and differences in various fields is appropriate. The step-by-step explanation given through the German traffic code text analysis was useful to organize the (at times) overwhelming and meticulous aspects of ontologies. Kuhn was critical and elaborate when discussing the limitations involved in textual language processes and future challenges of ways ontologies will be utilized in geographical space.

He argued for the representation of reality in geographical information to be prioritised less than what we do with that information. More specifically, how it is practical and what the user needs are. Even though I agree with the article, that practicality is a key factor in the development of textual ground, reality represented in geographic space should not be completely ignored. This is due to the lack of clarity to support the notion of the inability of ontologies be task-dependent. Hence, Chandarasekaran’s (1998) statement, “what kinds of things actually exist should not depend on what we want to do with that knowledge”. However, the various characteristics of reality of a domain which belong to a specific ontology (through identification and the written form), depends on the particular tasks the ontology is being built for (Chandarasekaran’s, 1998). Kuhn finds this to be critical to what can be achieved in practice. I believe a combination between practicality and reality would be most effective as the two are both substantial to ontological use in the geographic realm.

-henry miller

Ontologies: abstraction, imagination, existence

Friday, February 3rd, 2012

Being new to the field of ontology, I took a deep breath before starting to read what I automatically thought would be an obscure, existential article titled “Do mountains exist?” To my relief, it was much more than that. As a hiker, I first thought about my personal connection and idea behind mountains. Do mountains exist? Do I believe mountains exists? All of this is somewhat vague, leaving much room for interpretation; a question that will undoubtedly be answered with many, many other questions. Does this matter? Do all humans believe they exist? Or maybe just some? What is the construction of meaning behind determining their existence?

Arguably, this is a challenging field, and I believe Smith and Mark provided a helpful, in-depth explanation on the different dimensions and perspectives of ontology (focused on human thought and action). At the same time, the authors acknowledged their limitations as all concepts/issues pertaining to this topic could not possibly be addressed at length in the article. This was carried out by outlining the dichotomies of primary, and secondary theories; the former is grounded on an analytical approach, incomplete due to limitations in explanations, assuming common knowledge. The latter is comprised of folk beliefs, developed at different levels, with much diversity. This, in turn, is dependent on a specific culture or community, deeming secondary theory to be inconsistent.

I did find it interesting that a focus was made on primary theory, and the way it can be integrated with the “realm of science” (10) since it is the theory of the geographic domain (9). What happened to secondary theory? This makes me think of Ally_Nash’s comment of primary theory being objective and secondary theory being subjective. Is that what the authors thought as well and that is why the focus in the article is on primary theory? The authors attempt to merge philosophical and information systems approaches within a single framework (6), where “a complete ontology of the geospatial world would need to comprehend not only the common-sense world of primary theory but also the field-based ontologies that are used to model runoff and erosion” (18). Thus, I argue that due to the challenges behind this integration, primary theory is not objective. Furthermore, “maps do not represent mountains directly as objects with crisp boundaries” (12), where abstraction plays a critical role in our conceptualization of them. The similarities between Mount Everest and the Santa Barbara neighbourhood create a paradox that Smith and Mark only half solved, as both (mountain and neighbourhood) are “a product of socially established beliefs and habits” (14).

Although there is much work to be done, I admire the authors’ ambitious plan to find an ontological framework that can unify the perspectives of a vast number of fields to create a complete ontology of the geospatial world. Why not use abstraction and imagination to unite instead of divide these fields.

-henry miller

Re-think Mountains in GIS with Ontology

Friday, February 3rd, 2012

In GIS, mountains exist as a number of 0s and 1s. They may be stored in the hard disk as vectors, matrices, or even single values. By visualization, we extract those 0s and 1s from the storage, display them according to the user requirements, and label them as “mountains”. By this means, we admit that mountains exist physically in GIS research. But with ontology, which studies being or existence itself, it is quite hard to define what exactly a mountain is. By taking a look at the theories in geomorphology or hydrology, it is nearly impossible to find the starting and end of a mountain, and we can even challenge whether “mountain” is an appropriate name to describe the altitude of certain locations. But with information systems, ontology does not mainly deal with existence, but formalize the concepts under established logics or theories. To be more specific, in GIS, ontology helps us to clarify spatial information.

Let us get back to the “mountain” example in GIS. We need to give labels to most “mountains” a label for identification, such as the “Mont-Royal” on Google Maps. But is this label correct? What happened if we label it as “McGill Mountain” in another GIS? I think if we label it as “McGill Mountain”, someone can still recognize that mountain, at least most McGill students. But with ontology, we can easily figure out that “McGill Mountain” is equal to “Mont-Royal”, as they have the same feature in GIS.

One very interesting argument in the paper of Smith et al. 2003 is that they view environment modeling as field-based rather than object-based. But without objects, it is difficult to model filed itself. However, with ontology, the notion of “field” may be easier to conceptualize. But here comes the question: Does ontology differentiate with respect to the complexity of concepts?

–cyberinfrastructure

Developing New Geospatial Cyberinfrastructure with Ontology

Thursday, February 2nd, 2012

Nowadays, geospatial information can be collected with unprecedented speed from multiple sources, including a large body of geosensing systems, historical records, online GIS databases, and so on. On the other hand, user requests for the geospatial information are rapidly growing and the requests always involve distributed heterogeneous data processing. By distributed we mean data are stored or available at different servers, and by heterogeneous we mean data are kept with different format, and both features present great challenges in GIS research. As Kuhn et al. mentioned in their paper in 2001, most traditional geospatial information systems have concentrated on map contents rather than the actual user requirement, which leaves a gap between geospatial cyberinfrastructure and user needs.

Ontology has been proposed to help geospatial information extraction and sharing from the mentioned sources by Kuhn in 2001. The author suggests developing user-oriented GIS instead of map based systems, and using the notion of affordance to establish a hierarchical model of human activities. And their theories have been implemented with the German traffic code project, which has proven the success of utilizing ontology to build the new generation of geospatial cyberinfrastructure.

In 2010, Sieber et al. have built another ontology based geospatial cyberinfrastructure, which incorporates the China Biographical database, the McGill-Harvard-Yenching Library Ming Qing Women’s writing database and China Historical Geographical Information System. This geospatial cyberinfrastructure uses ontology to provide synthesized information about Chinese Women writers in Ming and Qing dynasty, their kinship, publication, and social communities’ information. Utilizing ontology in the design of geospatial cyberinfrastructure, we can enjoy the improvement in spatial knowledge access, discovery and sharing.

 

–cyberinfrastructure

The Recipe for A Problem-Solving Ontology

Wednesday, February 1st, 2012

Rather than taking an approach that questions the very nature of existence of certain human-created objects, Werner Kuhn’s article instead reads like an instruction manual for how to derive ontologies. Of particular use, he breaks down the actual phenomenon that need to be categorized into four useful bites that allow one to derive an ontology. First, one must create different types of objects; second, classes for these types to belong to; third, functions that each type might perform (an attribute); and, finally, algebraic axioms that define what exactly the actions are that a given type of object affords (618). Of course, the difficult part comes from actually trying to map out the linkages between all of these different types in a domain, then ordering them into a hierarchy which eventually should have “complete executable specifications in a functional language of activities and object classes” (628).

Having completed all this, however, one might wonder what the utility of engineering an ontology might be. The author, of course, provide us with an example of the German traffic rules as one example for understanding how setting up a hierarchical ontology might be useful. He arrives at this example because he believes “ontologies should be designed with a focus on human activities in geographical space” (614). This supposition is further supported by an argument that an accurate ontology should support human activity in space. The act of supporting works because an ontology can “capture knowledge about problem solving in the world” (616). While Werner’s interest lies only in knowledge about the world (616) rather solving problems, we might be able to extrapolate how developing an ontology for traffic rules might aid traffic authorities in constructing further rules or a system for adjudication in Germany.

In thinking about this approach to a problem-solving given what we’ve discussed in class so far, Werner’s recipe for how to construct an ontology appears to be closely akin to Agent Based Modeling. In effect, he believes in taking “types” which could, in some cases, be “agents,” and then figuring out their actions and the hierarchy of how these actions relate to each other. However, his method for constructing an ontology places no emphasis on the individual roles of agents (or types) as does ABM, but instead is more interested in each “type” for its own sake – and for where it sits in a network. In some ways, this reminds me more of a network analysis. Yet I can see how this approach lends itself to something like computer programming that already derives from and creates its own ontologies, and could be greatly aided by engineering a specific existence.
–ClimateNYC

What About the Eskimo?

Wednesday, February 1st, 2012

Barry Smith and David Mark provide an interesting article that gets at the ontology of geographic features in space, but we should carefully heed the authors discussion of how the definitions of geographic features vary from community to community. As “Madskiier” points out, the authors draw a distinction between primary theory (or a basic common-sense perception) of objects versus a secondary theory (that reflects diverse beliefs) that allows for processing of ideas like continuous fields.

It’s in discussing ideas like continuous fields that the authors make clear how little weight certain geographic terms have when looking at tangible reality. For example, continuous fields thought of only in terms of Western contour maps may not exist for cultures who don’t possess such a map-making tradition. Take another idea proposed by Smith and Mark: did Mt. Everest exist before the first human laid eyes upon it? Mt. Everest, itself a blob of elevation, the authors might say, existed long before maps. But the concept of Mt. Everest as a delineated object, perhaps, began existing only when it was drawn on a map. Yet the fact that primary theory allows mountains to be map-created “objects whose boundaries are marked by gradedness or vagueness” (13) doesn’t make mountains a reality. In fact, for those who haven’t seen Mt. Everest (or cultures, people or places for whom real, tangible “mountains do not exist” (13) ) it might be difficult to understand a map portraying one. Confused?

Consider the author’s discussion of geomorphology’s role as a science in studying landforms as supposed “natural” objects (17). Given the fact that landforms are not objects with discrete boundaries except insofar as humans define them, they do not exist except as a concept. By suggestion, the authors question whether or not geomorphology can truly be considered a science. Still confused?

Let’s try examining what I consider the main thesis of the author’s ontological arguments. Smith and Mark posit that many geographic features exist only insofar as human communities have “projected [] speech practices onto the surface of the Earth in such a way that they mark out certain territory with a certain shape and material constitution” (14). In effect, most of the geographic constructs we learn as children such as river, hill, valley, mountain – as the author points out – are simply human constructs of given features which we make into an object. As the authors note, “our common-sense beliefs are readily translated from one language to another, and judgements expressing such beliefs are marked by a widespread unforced agreement” (8). By learning certain concepts as babes, we are reinforcing them as reality (one might argue, perceptual reality), according to the authors.

So, does a mountain actually exist? Sure, insomuch as the term mountain describes what “actually exists in reality — or more precisely in some part of reality that is relevant to human perception and action” (11). Of course, this then begets the question of what is a mountain exactly to someone who has always lived on the flat plains of Kansas and never seen one firsthand? Or, better yet, could an Eskimo describe the category/object of landforms known as a desert after a lifetime living amongst snow and ice?

In this sense, Smith and Mark’s article stems from a long-line of critical theory first suggested by the French philosopher Jacques Derrida who originated the ideas of deconstructionism. Instead of asking about mountains, Derrida might have questioned whether the word “tree” means the same thing to a man from a dense, tropical jungle as a man from the Northern boreal forests of Canada. Since both men could correctly picture a tree but arrive at startlingly different objects, what does this say about the nature of the tree as an object? Or, in Mark and Smith’s terms, of less-defined objects such as mountains?
–ClimateNYC