Computers, Society, and Nature

I agree with sah about this article particularly with respect to the need to have task specific ontologies rather than a specific universal ontology of landforms in many cases. Those who study a mountain or require precise definitions of what a mountain is would require an ontology of landforms although they may be the only ones to use such an ontology. In class, it was mentioned that keeping spatial uncertainties present in the data was often very important in representing different views on intangible concepts such as disputed country boundaries. This same thinking can apply in terms of ontologies as well.

An ontology, to me, seems like a dictionary of the spatial meaning associated with a particular word. In this sense, and perhaps I have misinterpreted what an ontology is exactly, an ontology could have multiple definitions of a particular word and the user could select the correct definition for their purposes from the ontology. I compare this to different types of citations available on citation manager software. There are many different ways of representing crucial citation information and the user need only select the one they require. Why could this not apply an ontology of landforms?

-Outdoor Addict

GIS is known to have its pre-defined categories and way of being in the world. In fact, GIS has its own ontology, or even ontologies (as we can’t even decide if it is a tool or a science, we must have different ways of thinking about GIS). The question that must be asked is when is a parcel of land part of a river network, and how can this be represented in GIS today?

Smith and Mark outline in their paper that ways of thinking about mountains differ from culture to culture and language to language, this is also true when thinking of GIS.  GIS has a certain way of thinking about real-world phenomena that may differ from aboriginal perception of the same phenomena. Sieber once mentioned that in GIS, a river network is constrained to the line segment that represents it, but for aboriginals, the river might also continue as part of the land over which they portage their canoe.

The same can be said for other geographic phenomena, such as mountains, as in Smith and Mark’s article. Mountains are represented in reality as continuous landforms indicated by a steep elevation gradient, though the commons will identify a single mountain as an object, and even offer it a name.  Aboriginal peoples may attribute spiritual value to a mountain, or be the basis of their world view, as quoted in Smith and Mark. However, in GIS a mountain can be represented as a gridded digital elevation model, a point or a polygon.

Ontologies represent reality for a certain group, and also relates to GIS as a field.

-rsmithlal

The first thing I noticed was that Smith & Mark start with a much more philosophical definition of ontology, as being focused on describing “the constituents of reality…in a systematic way”, as opposed to Kuhn’s definition diving straight into specification of conceptualisations through language which bypasses the question of existence. It was interesting to see the two approaches – one from the domain based ontology and the other from the more holistic approach.

Smith and Mark provide a good overview of ontology, especially primary and secondary theory and the separation of the two. However, their actual suggestions on the future of geospatial ontology is quite scarce, apart from stressing the need for an all encompassing ontology that is general enough to be used in any scenario, but also able to be tweaked as well.

Kuhn on the other hand goes through the interesting process of creating an ontology, and puts more detail into concepts such as affordances. The methodology he goes through is interesting, but still very much dependent upon a textual source. The choice of that source is absolutely crucial – choosing a text in a certain language is probably already resulting in a loss in ‘resolution’ (if that term may be appropriate here), but it is after all, a domain specific ontology, in which case – why translate it to English in the first place (I must note at this point that I am definitely not an expert in the field of ontology).

What I would like to question though is whether or not having separate ontologies is necessarily a big problem. The ontologies used everyday I think are very much a cultural phenomenon in such that they are and should be flexible and malleable according to what humans do and the scale at which we are able to perceive things. In trying to create a formal ontology (an ontology that is unbiased and constant, independent of content), one is probably (as madskiier suggests) limiting the ability to express oneself. The nature of the world is dynamic and human knowledge increasing, so perhaps it is the nature of ontologies to grow, rather than be static. I do agree however that issues of translation and cataloguing are very reliant on ontology, but having separate domain based ontologies should still be the way to go, in order to preserve as much detail as possible.

Finally – what would a formal geographic ontology do for the imagination and communication? It may make the world a slightly more boring place

-Peck

*yes. cold desert environment.

The concept most fascinating to me in Kuhn`s paper was that of grounding which explained that the “claims of any domain theory need to be based on some observation in that domain”. This makes intuitive sense; one must know the domain in question and have some reference material to be able to begin to assemble the domain theory and an accompanying ontology. However, Kuhn refers to the “observation” as being “tangible” which may not always be the case. In Kuhn’s methodology, the basis for information is from a textual source compiled by experts. This assumes experts have the most appropriate grasp of how to assemble the concepts of a domain into some textual reference and then ontology. Yet this neglects two things: public participation and non-tangible reference material.

Public participation would be crucial in cultural studies where experts may not actually be of the culture they study and so may miss some cultural distinctions as a result of their own cultural influences. This reflects a need for the public to participate in defining concepts and objects as they see them and not only as these things are seen by experts or, as Kuhn mentioned, by the knowledge engineers often creating the ontologies.

Closely related to public participation are non-tangible references for a subject such as oral histories of some cultures. The history or stories told may remain relatively stable over time yet the words used to tell a particular history or story may change from telling to telling making it difficult to pin down exactly the meaning of a single concept in the story and to place this in an ontology.

-Outdoor Addict

In his article, Kuhn states that ontology design is performed by knowledge engineers that are not specialized in the domains they are designing ontologies for (Kuhn, 619; emphasis added).  Furthermore, he states that ontology design is carried out in consultation with domain experts through informal interviews, as well as incorporation of documents detailing the ontology requirements and existing databases on the subject of the domain(619).

I feel that these knowledge engineers would be able to create a more in depth and meaningful ontology if they were to actually undertake the grueling and likely expensive task of specializing in the domain that they with to create an new ontology for.  Borrowing from the field of Anthropology, you could think of this concept as a type of Ontological Fieldwork.  The premise of fieldwork is to be able to learn the subtle nuances of a culture, or in this case – a domain, through integration and participation in the life and events of a particular culture, or in our example, domain.  I view this current generation of Knowledge engineers as the equivalent of Armchair Anthropologists, those is in the pre-Boasian era of anthropology would study, and subsequently define, a culture based on explorer, missionary or colonial reports.  This of course lead to a the prevalence of ideas about cultures that were sometimes very far from reality.

In order for the effective and meaningful ontology design, I propose the following five steps to be undertaken by Knowledge Engineers before attempting to define a new ontology for a domain, over a period of at least a year.

1. Designers should identify and chose a mentor from one of the established researchers in the designers domain of interest.
2. Designers should study their mentor’s work, and subsequently explore the work of other established and up-and-coming researchers in the domain of interest.
3. Designers should attend or organize conferences or round tables designed to bring to light a collective picture of what the particularities of the domain of interest.
4. Designers should synthesize their findings and prepare a report, or ethnography, of the domain of interest.
5. Designers should commence work on designing an ontology to represent the domain of interest, taking into consideration nuances of the domain and other findings discovered during domain fieldwork.

In conclusion, I feel that ontology design would benefit from an added level of familiarity with the domain of interest. This would aid in illuminating meaningful nuances that may otherwise be overlooked when researching a domain using conventional, non-committal methods.

– rsmithlal

I believe an approach that is based on Gibson’s idea of affordances is already off to a good start because it places human experiences, namely activities/tasks, as priority. Theoretically, it offers us an opportunity to escape from the daunting task of defining the precise boundaries of a river or the exact place a mountain ends as objects of reality and focus on how features allow activities to be performed. However, more importantly, I think it fits in nicely with Heidegger’s idea that nature is fundamentally revealing. According to him, there are two ways of things come into being, either through “phusis”, where a thing rises from itself and retrieves back into itself (flower blossoming) or through “techne”, where a craftsmen brings a thing comes into being (carpenter makes a bench out of wood). In short, a feature in the world reveals a certain part of itself to us depending on the task we seek to accomplish.

I think the paper could have been improved if the authors first give an overview of existing methods of ontology construction,, whether object-oriented or field-oriented, and discuss their advantages/disadvantages. In particular, which aspects are poorly served by constructing a GIS ontology through an object-oriented approach and how does the approach based in affordances address these issues. This comparison would have helped to demonstrate the strengths of their method more clearly. One questions arose while thinking about ontologies that supported activities. If an activity-oriented ontology is based on certain activities, I wonder then, if an ontology can be constructed for a map designed for general purposes like Google Map or Google Earth?
– Ally_Nash

I think that Smith and Kuhn have different viewpoints of ontology. While Smith takes a “problem solving” approach, Kuhn approaches it with a more “critical theory” approach. That is, that Smith accepts the structure of the system and works within in it, while Kuhn attempts at a complete restructure. I personally found the Kuhn article a little more difficult to process, but I did find it interesting that he focuses on language and logic theory. I also noted that while Kuhn suggests that there are “Ontology designers” who appoint what may be included in GIS language and or experience. Contrary to this, Smith infers that our ontologies are subconscious and are created through our personal life experiences. He mentions (as other classmates have cited) that someone who is exposed to plains and fields might have a different impression of a mountain range.

ClimateNYC also mentions this, and I will add that not only would an Inuk have difficulty describing a desert, but it has been noted by many anthropologists that Inuit have an extended vocabulary for certain types of snow and ice. Professor Colin Scott (McGill) suggests they might have four, five or six separate words describing snow; all dependant on the temperature and moisture content etc.  This concept is foreign to even Canadians. I personally can only think of two; wet snow (slush) and normal snow (these days snow is almost non-existent and ice may soon replace the word snow).

Both authors take interesting standpoints. Personally, I like Smith’s perspective. I believe that Ontologies occur without conscious effort. They are formed through many years and experiences, each one contributing to a personal ontology of a thing or concept. I’m not sure that it is entirely correct to categorize and engineer a proper set of rules and defined ontologies.

Andrew “GIS” Funa

*I would prefer to not use the term Eskimo as it has been tied to certain negative connotations (that in itself may been ontological)

*I also will add that ClimateNYC infers that Inuit do not live in the desert, but just to be picky, I will add that they more often than not, do in fact live in a desert per se.

After reading the paper, I am still not very confused as to what the authors were arguing for. The different sub-headings seem independent from each other, which made their train of thought hard to follow. Often, I would find myself asking, “so how does that say about building an ontology of landforms?”

Before building such ontology, we as humans must first come to a clear understanding/philosophy of our relationship with our environment and the Earth. Although I am not sure if these terms can be equated, to me, primary theory sounds a lot like objectivism where things are deemed to exist upon being perceived and secondary theory sounds a lot of subjectivism where what exists depends on cultural believes and discourse. This dualistic way of understanding our relationship to nature is problematic. We cannot interpret our environment/space/geography as something that is external and separate from us when it is the very thing that allows us to experience ourselves. The French philosopher Merleau-Ponty has dedicated his life to constructing the Gestalt ontology. Naess has also extended this ontology in environmental philosophy. For me, the Gestalt ontology is the right direction we should be heading because it tries to marry the subjective-objective/mind-body duality in a way that the two views can remain distinct while remaining the same thing.

A gestalt “refers to a network of relationships whose various elements are mutually defining” (Diehm, 25). Gestalt is thus the smallest meaning chunk that we derive sense from. A good way to think about gestalt is by referring to a sculpture of a man holding a woman by Auguste Rodin. In one instance, the sculpture is called “Fugit Amor” and expresses the emotions of love, longing, and desire. In the second instance, the same sculpture appears on the door of “Dante’s Inferno” and expresses suffering and agony. Depending on the context and the meaning of the elements are changed completely.

I think an ontology of landforms should somehow encompass this way of relational thinking. Also, experience should be given the utmost importance because it is how we “know” the world and through it, how the world experiences itself (this is also way I like the idea of affordances). I would like to argue that the mountain does NOT exist before humans came along in the same way that it exists today after we have experience it because before experience a so called mountain is completely meaningless.

Diehm, C. (2006). Arne Naess and the Task of Gestalt Ontology. Environmental Ethics, 28(21-35). Retrieved on February 1st, 2012 from www.umweltethik.at/download.php?id=439

-Ally_Nash

In my last post, I questioned the need for developing a complete geographic ontology, as it perhaps marginalizes certain ways of knowing. Werner Kuhn—by illustrating how GIS can better support human activities through the use of ontologies, which specify and produce to user needs—has clarified many of my questions and concerns. For example, Kuhn forms a much stronger relationship between the inefficiencies of trying to access and use the many geographic datasets (of varying ontologies) with giving priority to human activities and encouraging efficient use.

One of the elements that I found to be interesting was how the relationship between the ontology designers and the subjects was highlighted. Designers and those who engage in field work are actively choosing what to include in their projects; they are not passive recorders, but instead engage with and impact their fields of work. I would like to note that while these decisions are explicit, choices are also being made at a sub-conscious level.

Khun suggests that a GIS’s purpose should guide how its language is determined. Because humans’ perception of the world is based off of the objects as well as actions, ontologies need to incorporate both. In the portage example, an ontology would not only include how a community perceives a path to be a part of body of water, but also how it uses it, thereby placing significant importance on the requirements of the community. By hierarchically structuring activities and objects (while paying very close attention to semantics), it had become clear to me now the way in which ontologies have the capability to include the varying perceptions of people’s environment while also enabling greater interoperability, which will increase efficiency and usability.

– jeremy

Smith and Mark posit that in order to address the incompatibilities that arise when data gatherers use varying terms and concepts, a complete geographic ontology needs to be developed. But this is a daunting task, as through engaging with the world, every person  develops a specific conceptualization of their environment. This conceptualization varies between members within a community (with regards to both primary and secondary theories), and often varies drastically between members of different communities. I would like to ask if the process of creating a complete ontology—as it seeks to facilitate understanding—in fact marginalizes certain ways of knowing.

As noted by Madskiier_JWong, the portage path example confronts our understanding of primary theory in that it is an ontology that is not ‘common sense’ to most of us. This way of perceiving the world, like Smith and Mark illustrate, is influenced by how the environment is being used. I have difficulty, however, in envisioning how this specific way of perceiving bodies of water could be incorporated into a universal geographic ontology. Could it be its own ontology portage ‘layer’ that can be combined with other perceptions of waterbodies? Or would it be incorporated into a larger set and adjusted to fit a norm?

In addition, Smith and Mark illustrate how the primary theories of certain geographical features may be underdeveloped for various communities. Mountains, for example, may be a fuzzy concept for those living on isolated flatlands. But to be beneficial to those living on isolated flatlands, does a geographic ontology need to include how others perceive mountains? In a human geography sense, it is arguable that the goal of creating a geographic ontology is so that it can be tailored and be of benefit to those it is being designed for. By going back to the portage example, it can be said that the concept of these paths being a part of waterbodies is in fact a geographic social construct (which may not be translatable to other communities) but representing this in contrast to other ways of imagining waterbodies is arguably its strength.

As for who a complete geographic ontology would be useful for, Mark and Smith note several examples in their conclusion: pilots, soldiers, scientists, hikers, firefighters, and naturalists. It is clear that creating a unified geographic language would be of benefit to these individuals, but this seems to be a top-down approach. In creating a geographic ontology, perhaps the focus should be on specifically tailoring it for the needs and values of the community at hand in order to understand and serve it—non-intertranslatables and all.

– jeremy

Combining thoughts of secondary theory as put forth by Smith and Mark and the topic of affordances put forth by Kuhn offers a complex evaluation of how different cultures use different objects. Kuhn quotes Gibson: “the affordances of the environment are what it offers the animal,” (617), which I follow with the assumption that environments may offer different “animals” different things, and spur various forms of activity.

In the beginning of his paper Kuhn poses the question, “what can be done to make geographical information more supportive of human activities?” (613). The process of creating ontologies includes grasping the actions from a text, then identifying the objects which afford them. I am curious to how much rigour is ensured during this process, and how much of the object-affordance relationship is left to assumption or distant inference. Kuhn notes that the “details of this procedure are also language-dependent” (622), insisting that some statements may be lost in translation. It is difficult for a person to try and examine the activities afforded to certain cultures by certain objects—since, as we know with knowledge of secondary theory, is not consistent across the globe.

Kuhn, Werner. “Ontologies in Support of Activities in Geographical Space.” International Journal of Geographical Information Science. 15.7 (2001): 613-631.

-sidewalk ballet

Kuhn’s article Ontologies in support of activities in geographical space was quite thought provoking for me.  As an overview of ontologies, where they are and where the author believes they could be, it was informative.  At the end of Kuhn’s piece, I found myself wondering two things.  First, I pondered the usefulness of ontologies–genuinely wondering how they are applied.  Iit would have been an interesting addition to their case study of an ontology of German traffic codes if they had also included the usefulness and applicability of where and how this could be used.  Particularly as this article emphasized the necessity of ontologies related to the task at hand, a knowledge of what the “task at hand” could be for such an ontology would have been a thorough and convincing conclusion.

More than that, though, it caused me to reflect on GIS as a whole, and the frustration that comes along with a software (tool/science/system?) that has the ability to include immense amounts of data and answer many questions.  As mentioned, Kuhn posits that “the tasks (or activities) to be supported by a GIS should determine the entities that are admitted to its languages”.  When using GIS as a tool, this is a mentality that should not be forgotten.  The task at hand is incredibly important.  GIS can accomplish many things, analyze interesting data, find solutions and, sometimes, create more problems.  As a user with a plethora of tools before you, it can be intimidating, or you may attempt to do too much.  With the geospatial and human world being incredibly vast and comprehensive, it seems like it should go without saying that decisions made, and tools–or in this case, languages–chosen should be inherent upon the context within which one is working.

–sah

Kuhn, Werner. “Ontologies in Support of Activities in Geographical Space.” International Journal of Geographical Information Science. 15.7 (2001): 613-631. Print.

An ontology of landforms is an interesting concept.  And this paper approached it in an interesting way, asking, “Do mountains exist?” moving from the various attributes of a mountain into the broader question of how might we characterize our landforms in general.  But I found Smith and Mark addressed multiple ideas without neatly pulling together their main argument and limitations to such an argument.  I was slightly confused as to precisely how they felt about some of their main points—can Primary Theory provide sufficient understanding to create a shared language for landforms, for example?

Upon reading this paper, I was not convinced.  I believe to create one ontology for landforms would be incredibly difficult, even if we are assuming the ontology would be used by a relatively homogeneous group of people (in terms of cultural belief systems/research parameters/etc).  I understand the necessity, as they state, of having a shared language for people such as researchers, pilots, soldiers, scientists, to name a few.  But how can we presume that all people will have the same needs of a landscape, and thus will benefit from the same ontology?  It begs the question “What ARE mountains… and why do we care?”  Well, we care because of the question we are asking—they are relevant to something of interest to us.  As Kuhn states in Ontologies in support of activities in geographical space, ontologies appear to be somewhat task specific.  And the way we define our landscapes is surely specific to how we will be using them?

It seems what they describe as “Primary Theory”, as I understood it, may not relate to understandings that are as general across the board as they necessarily make them out to be.  Perhaps a better knowledge of ontologies on the part of the reader (ie. myself) may have clarified aspects of this argument, or perhaps it is the argument itself.  But it is particularly ironic, in my mind, that in a paper discussing the benefits of a shared and commonly understood language, their ability to express and defend their position was lacking.

–sah

Smith, B, and D M. Mark. “Do Mountains Exist? Towards an Ontology of Landforms.” Environment and Planning. B, Planning & Design. 30 (2003): 411-428. Print.

Kuhn, Werner. “Ontologies in Support of Activities in Geographical Space.” International Journal of Geographical Information Science. 15.7 (2001): 613-631. Print.

Kuhn’s article is critical of existing ontologies as overly focussing on objects and their attributes. He details the development of a new ontology that embraces processes and actions in addition to objects. He advocates for the formalization of such a system to better represent the geo-spatial component of phenomena. As an example, he uses a German traffic code as a case study to demonstrate innovations in ontology structure.

A contentious point in this article was Kuhn’s assertion that teaching the programming, engineering, and ontology-generating tools to domain experts is useless because these tools are incompatible with the theories. I believe that familiarizing and engaging domain scientists in critical review of these tools (much like what this seminar class is doing) exposes the built-in assumptions and limitations of the ontology; an ontology that originates from computer science and information systems. Furthermore, contemporary neo-geographers and the geospatial web are blurring the distinction between geographers and programmers. Intuitive tools such as search engines and even programming languages that increasingly approximate natural language are emerging.

There is merit in Kuhn’s call for a shift of structuring data to better represent the geographic component of geographical data. However, I believe that it is entirely too abstract to create a new system from scratch. Much like how our understandings of things change over time, is it not possible to allow our ontologies to change organically and incrementally? Such a transition would allow for more recycling of collected data, more exchanges of dialogue, and is probably more feasible.  

– Madskiier_JWong