Computers, Society, and Nature

Ontologies are such an interesting and abstract field to me. In the lab I work in, there are many people who develop ontologies (some spatial, some not) and I always struggle to comprehend what they are or how I could ever explain them to someone. They seem to be classification systems or ways of understanding trends in different types of data. For example, one project is based on looking through blog posts on vaccines and classifying the content as “pro-vaccine”, “anti-vaccine” or neutral. The idea of how you would do this was completely abstract to me before I read this paper, now I can see how it fits into some of these concepts. The ‘sentiment’ of the blog is similar to a “secondary theory” relating to the content. As with many analytical models, ontologies to capture spatial trends are more complicated than their aspatial counterparts, but also raises a whole new set of interesting challenges (e.g., issues of scale). Looking forward to the presentation tomorrow and maybe finally  really understanding ontologies!

-Kathryn

[PS My spell-check in firefox seems to thing “ontologies’ isn’t a word and wants to change it to gerontologist…]

Nelson’s article gives a thorough overview of spatial statistics through a synthesis of literature and surveying of professionals in the field. The article is well structured as Nelson walks the reader through the different sections. From this article, spatial statistics can be linked the past topics that we’ve studied in GIScience, such as the importance of good user-centred GUIs and wide distribution of applications on the web and data visualisation.

Nelson has a subsection entitled “Geography as the Home for Spatial Analysis” where she situates spatial analysis within geography. She comments on the trend of certain subjects migrating into different disciplines (or forging their own) as geographers give up leadership. If the growing fields within geography leave the discipline, what do we have left? Are geographers equipped to meet the demands of the growing fields? Nelson continues on to acknowledge how geographers are not trained to think mathmetically, statistically, nor computationally — strains of thought which are required for spatial stats. She raises questions on to what extent spatial stats should be involved in geography’s curriculums. I think McGill does a good job with our two required stats courses, but I would like to see more application of statistical methods in other courses.

Spatial statistical analysis needs geographers — maybe not to perform the analysis, but for spatial interpretation. Geographers need spatial analysis to increase rigour in our studies and validity as a department.

Interestingly, this blog post calls for geographers wanting to become spatial statisticans to round out their education with a math or stats degree. It takes more than just geography.

-sidewalkballet

PS: For future thought– Nelson says, “data are increasingly being viewed as public properties” (p. 86)… hmmm…

As with many of the papers in this class, the topics presented are still extremely relevant to the field of GIS, however we have made leaps and bounds in terms of technology since it was written (1992 in this case).  Computing power and the development of appropriate algorithms have allowed GIS analysts to drastically improve the so called manipulation, exploration and confirmation processes brought forth by Anselin and Getis.  While I have only been familiar with a program such as ArcGIS for a few years, I would argue that the spatial analysis capabilities have drastically improved since the 90s.  It is obvious that GIS is no longer about the display and visualization of spatial data, as the ability to perform exploratory and confirmatory analysis has become the norm.  These sorts of procedures have become more “automated”, per se, and allow for more “plug it and chug it” methods to spatial analysis.

That being said, the authors bring up a vital point by saying that in some cases, “better theoretical notions may be needed.”  To me, this is essentially a warning to GIS analysts, telling us not to rely solely on whatever new algorithms or spatial analyst tools may be needed.  When one is working with the massive complexity of spatial data that is at our fingertips today, it is imperative that we are familiar with the data itself.  We must still predict what sorts of patterns we may see.  If the exploratory process unveils some sort of new model of our environment, we need to know why that is so.  Otherwise, we reach a point where the user is no longer relevant, which will be detrimental.  So, yes, we have made great progress in the use of spatial geostatistics.  However, we must be careful how far we take this and always be conscious of the types of decisions we make when analyzing spatial data.

-Geogman15

Source: xkcd

Elwood reviews areas of GIScience that are crying out for advancements in the wake of new, Web 2.0 technologies and the flood of big data coming with them. More and more user-generated content is being created every day, and a great deal of it is embedded with explicit or implicit geospatial information.  Two key stumbling blocks to harnessing these new sources of data are the heterogeneous, unstandardized nature of Web 2.0 content, and the qualitative nature of spatial knowledge within the content.

Heterogeneity of geospatial data on the web is an acute issue: because content is user-generated, the range of standards and conventions used in terms of communicating information is huge.  When you factor in the equally diverse, and growing, number of Web 2.0 platforms and services, things are even more daunting.  If there’s a certain class of content you’re interested in analyzing, how do you make a structure out of what is mostly unstructured text? How do you standardize when the number of standards is longer than your arm? As the XKCD comic suggests, creating a new standard us typically an exercise in futility.

The qualitative nature of spatial information and meaning is another stumbling block in the use of data from the Geoweb. As an example of this, suppose you want to geocode a (location-disabled! Ha!) tweet of mine that says “Home, finally :)”. Even if you have detailed biographical and address information about me, you will still encounter ambiguities. By “Home”, do I mean my house? My neighbourhood? My hometown? My residence while I’m at school? My parents’ region of origin where my extended family lives? How do you even begin to automate the processing of such context-dependent information? And is there any existing spatial data model that can effectively represent the concept of home?

It is often said that while Geoweb 2.0 technologies are accessible, engaging and intuitive for a much greater number of people than traditional GIS has ever been, analysis tools for this exciting new medium lag behind GIS.  This is in no small part due to the issues associated with organizing the web of heterogeneous, qualitative data that these new tools are producing.  Though this challenge looms large, Elwood is optimistic that GIScience can rise to the occasion if we take a multidisciplinary approach.

-FischbobGeo

Lanter and Essinger write about the cutting edge of graphical user interface (GUI) design for GIS in 1991.  This was an era where the command prompt was still in vogue as a user interface and computer use was still unwieldy for many people.  It was also a time of change as more GUI-based software such as Microsoft Windows was gaining popularity.

Though the technology Lanter and Essinger profile has aged considerably since publication of this piece, the fundamental principles underpinning systems design and the system-user interface are still as relevant as ever. The authors discuss the need to move from a systems-centric design paradigm where operation of the software simply reflects the underlying algorithms and processes as directly as possible, to a user-centered one that gels with the user’s own mental models of the tasks they are doing.  When the latter succeeds, controls and functions are more intuitive for the user, who then does not need to be bogged down by excessive documentation or cryptic commands that only make sense to the developer.

An important point concerning the user-systems interface relationship is that it is a two-way relationship. Not only should system interface design be influenced by the user’s mental models, but those same mental models can be changed by interaction with the software, especially when the UI is set up in such a way to allow learning through exploration.  Most of my computer savviness stems from just being able to explore and mess around with things on Windows 95 and Macintosh machines starting at a young age.  My expertise with touchscreen devices, on the other hand, is far less developed. I am terrified of something happening to my Android phone, because I don’t know enough about the underlying system to be able to diagnose and solve problems as I can do with a more traditional PC environment—at 21 years old, I’m already set in my ways!

This has important implications for software developers who wish to advance human-computer interaction but find themselves faced with a generation of users most comfortable with the keyboard/mouse.  UI advances must be implemented and deployed incrementally so this cohort’s mental models have the opportunity to adjust. Unless a very ‘intuitive’ design is found, too-radical changes are bound to fail and be looked back on as being ahead of their time.

-FischbobGeo

My initial reaction was to question how GUI’s make people even more distant with computer technology. If GUIs are made so precisely such that the underlying technology is completely hidden from the users, then you can run into problems where users click randomly without really understanding what the tool is doing.  But after reading further down, it gets more complicated than just button mashing, hiding algorithms, and hiding all the techy things the user never signed up for  (obviously). At the core of topic was to create a successful user centered interface – a marriage of what the users knowledge and the process maps/models in their mind and how the tool can adapt to that to further add/influence the user’s model. A great concept that Lanter argues will reduce the necessary documentation, software support and unnecessary brain space traditional GUIs demand. However, for which user should the program be created for? How can the program be created for a beginner who is just learning the tools and concepts necessary to navigate through the problem, and expert users who would like the ability to create custom functions? Or what about people to visualize conceptualize information different. Like Professor Sengupta’s analogy about road directions  (western countries may be more familiar with “next left, next right, continue easterly” type of directions while Asian countries are more familiar with landmark directions), this also applies to developing GUIs that tailor to the dominant norms in a particular society. For instance, some Asian societies traditionally read from right to left thus having menu bars and text left justified in less intuitive for them and forces to adapt.

A prime example of user interface GUI within the geography realm is the upgrade from ENVI Classic and the new ENVI 5. The designers of ENVI 5 definitely made the connection that those using ENVI are very likely to be using ArcGIS, and gave the user interface a very familiar ArcGIS feel.  For the users that this applies to, I think that this does reduce the steep learning curve of the tool, while perhaps enhancing the notion of the interactiveness of these two tools.

-tranv

This paper by Poole presented some very cool and interesting technologies that allow us to follow eye movements and relate them to visual interpretation.  While this technology seems very “out there” to me at times, there are definitely some practical applications that I can think of.

With the increased amount of information and data available to us, we must filter out what we don’t want to see. With the advent of digital earths and virtual environments being the medium through which we view spatial information (business locations, road networks etc…), there may be a point where the amount of information is just too much to handle for one person.  Using eye tracking technology, experts could figure out how to best present certain types of spatial data (e.g. houses vs. business).  Then, as a user we could pick and choose which data we want to see.

For example, say I want to look at all the business in an area, filtering out residential and other land uses.  Knowing how our eyes react to different representations, the interface could “highlight” the business locations.   Theoretically, it would save the user time from having to scan through all the information they don’t want to see.

Through these kinds of things, eye tracking could extremely improve the way virtual worlds are presented.  Oddly, it may even allow us to view parts of the real world in a more “efficient” manner than in reality.

-Geogman15

The challenges posed by MacEachren and Kraak are manifold, but they effectively outline the path that geovisualization has taken since the paper was written in 2000.  The increased use of the World Wide Web, increases in hardware and software technology as well as focuses on relevant theory has solidified geovisualization as an extremely useful and widespread field.  The huge increase in multi-dimensional, dynamic and interactive visual maps has allowed for a broader visualization and analysis of geospatial data.

One aspect that I think still presents a challenge today is the incorporation of group work and the extension of tools to all people that need it.  This brings us back to the notion of the digital divide that exists in our world.  As western based academics, pushing to the frontier of geovisualization is in line with what we already know and understand.  However, in parts of the world where the internet may not be as accessible, some of these newer features may not be as well understood.  As a result, a lot of new tools and programs may need to be adapted to certain people’s cognitive abilities.  Additionally, everyone interprets the space around them in a unique manner, calling for an in depth analysis of how geovisualization could help them specifically.

Despite this, I believe we have made huge leaps in geovisualization since this paper has been written.  Citizens can use many different applications that were only ideas to experts a few years ago.  Still, one of the main issues today is extending the progress that has been made to all corners of the globe.

-Geogman15