Computers, Society, and Nature

Dunn & Newton’s article “Optimal Routes in GIS and Emergency Planning Applications” (1992) heavily discusses the mathematics behind the Djikstra algorithm & its spin-off, the “out-of-kilter” algorithm, and the “out-of-kilter” algorithm’s use in early ‘90s GISoftware and on early ‘90s computers.

The “out of kilter algorithm” that diverts in multiple paths for an increased flow from one node to another, like an increase in traffic in emergency evacuation. I would have liked some more information from this article on the possible uses of network analysis for everyday people, but I agree this could have been difficult as personal GISystem use did not really exist then like it does today. The network analysis that Dunn & Newton discuss uses set points with available road networks for its running example, but they could have considered a world using network analysis that could rely on (unscrambled, post-2000s) GPS & constant refreshing. They briefly mention that some emergency vehicles have on-board navigation systems, which infers that they had the capability to discuss GPS & network analysis further, but did the inaccuracy of GPS at the time affect the emergency vehicles? Also, without these systems, a user would have to start from a set route and end at a set route and be limited to analyzing within a specific area that 1) their computer could hold and 2) their data was collected on, and on-the-fly adjustments (commonplace now) could not occur without extensive coordination. 

I am looking forward to learning more about current uses & future advancements, especially now that GISoftware isn’t just reserved for highly specialized people as it was in 1992, and that computers are faster (and that cloud computing, (more) accurate GPS, and mobile devices exist)!

The broad goal of knowledge discovery in databases (KDD) is, fittingly, to construct knowledge from large spatial database systems (SDBS). This goal is achieved via spatial data mining methods (algorithms) which are used to automate KDD tasks (e.g. detection of classes, dependencies, anomalies). Without a fuller understanding of the field at present, it is hard to judge how comprehensive an approach is outlined in Ester et al’s (1997) paper.

The authors underline the distinguishing characteristics of spatial databases; namely, the assumption that an object’s attributes may be influenced by the attributes of its neighbours (Tobler). These assumptions motivate the development of techniques and algorithms which automate the identification and extraction of spatial relationships. For instance, a simple classification task can be executed by algorithms that group objects based on the value of their attributes. The authors present a spatial extension of this approach, by incorporating not only an object’s attributes, but also those of its neighbours, allowing for greater insight into spatially classified sets of objects within a SDBS.

Contrasting with last week’s topic, the approach to knowledge extraction here emphasises automation. The goal is to construct basic rules that can efficiently manipulate and evaluate large datasets to detect meaningful, previously unknown information. Certainly, these techniques have been invaluable for pre-processing, transforming, mining and analysing large databases. In light of recent advances, it would be interesting to revisit these techniques to assess whether new spatial data mining methods are more effective for guessing or learning patterns that may be interpreted as meaningful, and to consider the theoretical limits of these approaches (if they exist).
-slumley

Ester et al. (1997) propose basic operations used for knowledge discovery in databases (KDD) for spatial database systems. They do so with an emphasis on the utility of neighbourhood graphs and neighbourhood indices for KDD. When the programming language began to bleed into the article it was clear that maybe some of the finer points would be lost on me. I was reminded of the discussion of whether or not it’s critical that every concept in GIScience is accessible to every GIS user. I’m convinced that in order for GIS users to practice critical reflexivity in their use of queries within a database, they ultimately need to understand the fundamentals of the operations they utilize. After making it through the article, I can say that Ester et al. could explain these principles to a broader audience reasonably well. I’ll have to echo the sentiments of previous posts that it would have been interesting to see more discussion of this, but perhaps it’s beyond the scope of this article.

Maybe it’s because we’re now into our 9th week of GIScience discourse, but I felt that the authors did a particularly good job of situating spatial data mining–which, despite its name, might appear more closely related to the field of computer science at a glance–within the realm of GIScience. Tobler’s Law even makes an appearance on page 4! It’s an interesting thought that GIScientists might have more to contribute to computation beyond the handling of explicitly spatial data. For instance, Ester et al. point to spatial concept hierarchies that can be applied to both spatial and non-spatial attributes. You can imagine how spatial association rules conceived by spatial scientists might then lend themselves the handling of non-spatial data as well.

Solcum et al. (2001) detailed emergent research themes in geovisualization circa 2001. The authors advocate for an interdisciplinary approach incorporating cognitive and usability engineering principles to address challenges concerning immersion and collaborative visualization. It was striking to realize how frequently I’ve brushed over the finer points made by the authors over the year and change I’ve spent submitting GIS assignments.I feel that so many without technical GIS training are inclined to conceptualize the discipline as “mapmaking.” In contrast it’s interesting how little time is spent on more nuanced cartographic considerations in introductory courses. The article made for a good introduction for engaging more meaningfully with what’s quite literally right under my nose.

Even though the article was presumably written before the release of Google Earth (B.G.E.?) it would appear that most of their discussion concerning emergent research themes is relatively robust–even if perhaps some of their associated challenges have since been addressed. For instance, I am not sure of what more could be said about maintaining orientation in explicitly geographic visual environments, but I would interested to learn more about how one would handle orientation in alternative spatial environments. Particularly such that would be immersive enough that would enable the type of cognition that we use in handling the real world. Moreover, I wonder how the ubiquity of Google Earth alone has propelled the topic of cognition and usability in geovisualization.

In the article “Cognitive and Usability Issues in Geovisualization”, Slocum et al.  discussed a need for maps or visualization tools to be conceptualized as  composed of both theory-driven design as well as usability engineering. The theory will describe how people think about maps with preconceived ideas about symbology, colour, layout, and representation. I thought it was super interesting to find out that different languages perceive different geographic features differently (as they noted, English and French perceive lake and pond differently), as well as other cultures perceive other colours differently. Along with the other more well-known differences between people, like sex, age, and sensory abilities, these can change ways that people view or look at maps. “Masculinist” has long been a term used in critiquing mapmaking and geovisualization, as the representations often favor a “God’s-eye”, flaneur-ish approach rather than other views. Geovisualization, particularly 3D visualization, may have the ability to change this. I think it would be interesting to revisit the emerging trends and (formerly) current standards that the authors review, to see where this representation has changed and where they envision it going to. I am not very caught up on the progress of the world of AI in geoviz, but the world of GIS has certainly changed with handheld digital maps like Google Maps or OSM, and even some of the “maps that change in real-time” has changed drastically (for example, manifested in Snapchat’s Snap Map).

It would also be interesting to learn who follows the methodologies laid out by Slocum et al. Though they do think more rationally about inclusivity, it doesn’t seem to be entirely all encompassing (ie. asking different groups of people what they like and don’t like about a geoviz and then working in the public’s comments). Further, do people actually use this advice? In video games, many use a lot of geoviz techniques to make the game world more realistic. Do game developers follow these trends? And more importantly for research and academic purposes, have the game developers shared their techniques of bettering geoviz with other industry professionals (like reducing “cyber-sickness” (6), or color choice, etc.)?

This paper was a comprehensive review of the current issues in geovisualization,  with a focus on legibility, and user-centered cognition, and features of experiential VR. They provided a number of conclusions based on their review of the current literature and the state of existing technology, and one of their recommendations was for more research in the cognitive sciences to identify the best methods for visualizing data to ensure maximum comfort and comprehension. I would have liked some more specific recommendations as to areas within cognitive psychology, or especially pertinent methods, which they believe would be useful for this field of geographic visualization.

Early in the paper, the authors note that if we develop”theories of how humans create and utilize mental representations of the environment, then we can minimize the need for user testing of specific geovisualization methods.” But I think that even if we formulate theories about how humans internalize, process, and store geographic information, this does not preclude the necessity of user testing specific methods. As they discuss later in their paper, there are considerable individual differences and a high level of specificity for each kind of VR representation, so user testing in each instance seems like a crucial step.

The authors commented on the paucity of publications related to 3D cmapping in comparison with the prominence of new softwares, and this seemed to be another manifestation of the GIS tool/science discussion we have been revisiting in class. In the sense that, should geo-VR and visualization be considered a tool or a science in itself? This usually leads to a further discussion of what constitutes a science.

One of the types of collaborative geo-visualization that the authors mention is the different-place, same-time scenario. I thought the most obvious instance of this is the multi-player  video games that people play over the web in real time, with people often located in entirely different countries. These games can be quite immersive and require considerable synchronization of timing, representation, and events, from multiple perspectives.

There is quite a positivist sentiment permeating this paper as to the power and potential of geo-visualization and in the discussion of education, the authors state that “we know so little about the ways in which children’s developing spatial abilities can be enabled through visual representations-” but whether or not this technology “enables” improved spatial abilities has not been established- they could have neutral or even deleterious effects on the development of spatial visualization and navigation skills.

-FutureSpock

I thought this article was really interesting, as I didn’t know too much about UAVs and their uses (and access to their imagery). The article covered the uses of (and constraints) of imagery from UAVs, and offered a refined list of online or free sources of imagery and editing software to process these images or videos.

This relates to a discussion which we often discuss in class, about the availability of funding and funding’s ability to steer the direction of research. Open data, like Open Aerial Map and others, are incredibly useful for those who may have often-imaged areas to research, and it helps to reduce the cost of conducting research while still allowing knowledge-production to occur. It seems possible that with the costs of UAVs decreasing, this may help remote sensing knowledge production to continue without a lot of funding (if the area is accessible, if the researcher has access to an UAV and the knowledge to use it properly, and access to the sensor they need, if the resolution works for their purposes, etc.).

Further, there is also software to process these images for free (of which I was previously not aware). Though there were some listed that were pay-per-use, it was really interesting to learn that free programs exist for RS processing. I have yet to open these sources and investigate them myself, but they seem like a good step towards a lower threshold for learning about and conducting remote sensing or even just aerial photography processing. Granted, there is always a worry that VGI/PGIS will be inaccurate due to the low threshold that “non-experts” can contribute to these sites, but I think that for basic use or for use in a project where higher inaccuracy/coarseness in data can be afforded, it’s a good resource. Further, I think these programs should be used more in an educational environment to avoid reliance on a specific company and give students a greater breadth in learning about different software packages’ capabilities other than the name-brand or industry standard (see: ESRI).

I think the article draws an interesting comparison between volunteered drone imagery (VDI) and other forms of volunteered geographic information (VGI). The “rise of the amateur” in most other VGI applications was really enabled by the spread of personal computers. It’s difficult for me to envision a world, at least in the near future, where personal UAVs become similarly prolific. The authors note that even for those that would purchase UAVs for enjoyment, there is still a technical barrier that would prevent less knowledgeable users from contributing. I imagine that it will be awhile before VDI contributors begin to resemble “amateurs” in the way many VGI contributors do more broadly. There’s probably an interesting discussion to be had about how different motivations behind contributing VDI and other types of VGI might affect concerns about data quality. I would be inclined to posit that VDI contributors have more professional expertise than that of the greater VGI community, perhaps making VDI less vulnerable to issues of  credibility and vandalism. However, it’s conceivable that fewer users with the appropriate technical expertise would give rise to less power of the crowd to catch and rectify errors.

I think another important distinction between VDI and other VGI projects like OSM is that many remote sensing contributions are likely less interpretive. For instance, an OSM contributor might delineate a boundary between wetland and forest from aerial imagery through tags. It would appear–based on my limited experience with remote sensing–the collection and contribution of most VDI would precedes these interpretive steps, so naturally there would be different ways that one would go about addressing accuracy and precision. Of course, if the definition of VDI were to include remote sensing derivative like classifications and DEMs (per the “UAV Mapping Workflow) address challenges associated with interpretation are unavoidable.

I found this article super interesting, as it discussed the complexities of visualizing social ties over space. I had never thought about 1) the fact that most relationships are thought about at least a little bit geographically but geographic visualization systems have not been able to visualize these dynamics (because they are, as Andris notes, “crude”) and 2) all this data from varying sources being used (possibly) to piece these things together that one takes for granted.

It is also interesting that this is an up-and-coming concept. It did not seem as though Andris was expressly studying social media platforms as much as she was studying the relationships between people expressed through data (through paper, telephone, email, social media platforms). It is incredibly surprising that this is not picked up further. In the big data studies that I have read, it seems as though most studies focus on numbers, the “butts in seats” raw numbers, and avoids connecting this information to the actual users themselves; Andris’s discussions of the nuances of geo-visualizing social networks actually tried to tie people’s digital presences back down to their humanity, which I found refreshing.

Further discussion of the geoviz nuances and making this discussion more “mainstream” in GIScience would be incredibly useful in urban planning or other disciplines that try to bring people together and make life easier for the people they serve. It seems to me that these visualizations have the potential to help decision-makers to do things better by actually seeing these numbers as real people’s real lives and livelihoods– though I do agree/worry with the classmate who posted earlier that this information could also be used nefariously.

Ridal et al. (2010) synthesize sociological and geographic techniques to investigate gang related activity in an eastern policing district of Los Angeles, California. The article challenged my conception of space and its role in GIScience. I can appreciate how social networks can be “spatialized” in relation to geographic information, but it was interesting how the language surrounding social networks themselves mirrors the language used more broadly in GIScience. How entities can be associated with a “location” in social or network space made me consider how other concepts I wouldn’t consider to be inherently spatial might be framed in a spatial context.

Their discussion of “spatial fetishization” really resonate with me, particularly in my experiences outside the Department of Geography. Mentioning my minor program to a group project member might prompt enthusiasm about the idea of “doing GIS,” and how we could incorporate it into our assignment. This could be especially true in the School of Environment, where GIS is touted as a uniquely hirable skill in a program that might otherwise emphasize theory over practice, but more generally I think the proliferation of GIS tools beyond the field of geography has the potential to generate excitement about exploring the spatial dimensions of a topic in a way that lacks nuance. The cluster analysis exercise was a good example of how a purely spatial approach alone might oversimplify a multidimensional question.

The binary classification of gang relationships being rivalrous or non-rivalrous seemed to be a little reductive. I was hoping the authors would explore further how one could address the evolution of social networks over time, but I agree this might be beyond the scope of the paper.

The author proposes a stronger embedding of SN systems in geographic space. In their framework, a node (geospatial agent) in a SN has its geolocation information formalised in the concept of an ‘anthropospace’. Unlike previous descriptions of human movement (e.g. life paths, anchor points), the anthropospace is a fluid concept which can refer to points, lines, areas, probability clouds associated with a social agent’s ‘activities’. I think this terminology is compelling in its universality, but may (as emphasised by the author) present challenges in a GIS setting. For instance, how should GIS deal with nodes that have different types/ scales of anthropospace?

I’m not convinced Andris’s system represents a definitive framework for resolving SN and GIS, but it does offer significant insights and examples. Further work would be necessary to persuade readers that the suggested typologies are exhaustive, non-arbitrary, and widely useful. Would a technical fix (making GIS software more SN compatible) solve this problem? I agree with the author that a conceptual understanding also needs to be advanced.
-slumley

Maybe the most challenging aspect of the paper–aside from the technical jargon–was trying to connect cyberinfrastructure concepts with my own experiences using GIS. Wang and Armstrong describe an approach to GIScience topics that I have never properly confronted. For instance, if someone were to ask me about the “crux” of inverse distance weighting, I would probably mention the definition of the relationship between interpolated values and surrounding points, or the selection of an appropriate output resolution. Increasing the near neighbour search efficiency is not immediately called to mind. If this paper had been my introduction to the study of GIScience, I would have likely begun with a different opinion about it’s place in the field of geography over computer science. Anyway, I can appreciate the appeal to the authors’ target audience through the conceptualization of “spatial computation domains” as spectral bands.

It’s true that some of the finer points may have been beyond my understanding. Still, the question on my mind as I read was whether or not my understanding was really necessary. As an end user, is my comprehension of the underlying cyberinfrastructure critical for me to evaluate the suitability of an algorithm? If it adds no additional source of bias or uncertainty, maybe not. Of course this is a big “if” as I am not confident in my ability to identify any such sources should they exist. In any case, it’s conceivable that in the age of big data GIS practitioners will need increasingly sophisticated tools to accommodate unprecedented data volume and reduce processing time.