Computers, Society, and Nature

This article introduced the potential costs brought by open data. Two categories of costs are identified in this article: direct cost and indirect cost. The direct cost of open data can be understood as collecting raw data, preparing data, and maintain the uploaded data. Then the author listed four indirect costs of open data: 1) issue aroused by citizen participation and engagement, 2) difficulty in reaching a sole standard due to unevenness of data provision, 3) tension between government expense and private sector use, and 4) the privatization of open data in private sectors.

I am very interested in the privacy issue in open data due to previous experience with crime datasets. In one previous project, I worked with Toronto Crime Datasets from its government open data portal, and I found out in the data acknowledgement that those points data have all been calculated by certain algorithm so that they won’t necessarily represent the true location of each events. Since this data is available to everyone who have access to the internet, I understand that this calculation is for privacy protection. This little change did not impact much to my project. However, what if some researches really need these kinds of information? Should the government giving out the raw data despite the privacy issue? What rationale should they use in terms of considering giving out sensitive datasets? To me, this is a dilemma of open data should be open or not, and the rationale of this question might also differ between different area or territories.

Open data has become a big movement in local governments. This article raises concerns over the costs incurred in the process of government data provision. The idea that making government data freely accessible – especially when geospatial data is involved – would create direct and indirect costs.

The authors suggest that direct costs must include considerations of individual privacy and confidentiality. Indeed, privacy protection may create direct costs, but government officials must ensure that all open data respects and only discloses information that cannot be attached to individuals. For instance, journey data is being used in a variety of ways to create and improve geospatial data products and to deliver services to uses. The journeys people take can be used to infer where they live, where they work, where they shop. If individuals become unwilling to share movement data, then this will impact the ability for that data to be used in ways that create economic and social benefits.

Besides direct costs, Johnson et al. (2017) identify four areas where the provision of open geospatial data can generate unforeseen expenses. They indicate that the private sector pushes for the release of “high value” datasets to develop their commercial products or services. This could divert governments’ attention from “low value” data. However, note that high-value data could also have a significant impact on citizens. People are taking advantage of applications that made use of open data. Transit commuters know how long they’ll be waiting for a ride. Drivers easily find parking close to where they want to travel to. Renters get detailed information about crime and school for any address. The information that developer access to inform these applications come directly from high-value datasets.

One way to reduce costs is to limit what data sets are published. Public officials tend to focus on the number of datasets they release rather than on the effect of releasing high-quality sets of data. Cities should do a careful analysis of which datasets have the most impact, both in terms of social and economic benefits, so as to avoid hidden costs.

This article framed the direct and indirect costs of geospatial open data provision, with the main focus on the four types of indirect costs. I found this article very thought-provoking because we often think of the benefits provided by open data whereas neglecting the pitfalls that it brings.

One point that particularly interests me was the data literacy issue. The article points out that there exist a number of barriers for users so that even though the data is open there is no guarantee of its use. Similarly, Janssen et al.’s (2012) article argues that these barriers pose the risk that open data is only publicized data in name but is still private in practice. Two points that I want to make here. First, while I understand the advocacy for better data quality and standardized data format, what I want to hear more about is that why does it matter for both researchers and the public to be able to use the data. One could argue that not many people would actually care and researchers are the group that those data meant for. Is public engagement in using and interpreting the open data instinctively good, or does it provide greater returns for the public? I think this could be better clarified here. Second, I’m curious about if VGI or crowdsourcing data belongs to the category of open data.  Dose the costs discussed in the article still apply to VGI and crowdsourcing data? It’s clear that some direct costs such as the cost of data collection could be avoided, but it seems to me that some other issues such as privacy and data quality could be intensified. I think this a question that worth to be discussed.

This article talks about scale, which is a key information in spatial variation. Scale as a concept has a lot of definitions. Outside the field of geography, it can be used to describe the size or extent of certain event or process. However, scale often referred to as cartographic scale in geography. In physical geography, scale is often represented using ratio value (e.g. 1:1000); while in human geography, scale can often be represented using units such as division, city, and province. In either study, defining a suitable scale is really important. If the original dataset is not appropriately scaled, it would be very useful if the dataset and easily be rescaled. In my past project, I had working on two different census dataset with different unit. When using them separately, they all worked very well. However, it took me a lot of efforts to co-register this two datasets in order to use them simultaneously. This makes me wonder if we were trying to build a dataset, what would be the criteria of a suitable spatial scale. In other words, how do we characterize suitable in this case.

This article then brought up the concept of spatial variation. It was first introduced to me as a part of point pattern analysis. Kriging is also introduced as a method to estimate and then interpolate missing data values. However, it has a ‘smoothing’ problem. Is there any better way to solve this problem since I noticed that this article was published in 2000, which is very early.  And with the extensive development of computer science algorithm, is there any new interpolation technology that can minimize this problem?

In this article, the authors highlighted the importance of spatial scales  in geography and the often need of rescaling data for multi-scale analysis. They raised the problem that due to the scale-dependent spatial variations, this rescaling process is very difficult. To solve this problem, they proposed some geostatiatical approaches for modelling the spatial dependence (variogram) and to predict the effects of rescaling (generalization). They suggested that this should be the first step for addressing scale-dependence problem.

The discussion of scale-independence processes reminds me of Anderson’s (2018) article “Biodiversity Monitoring, Earth Observations and the Ecology of Scale”, in which author discussed the need of muti-scale modelling and multi-scale mapping for biodiversity monitoring. The author explained that since the biodiversity pattern is driven by multiple ecological processes that act across different scale, there is no single best scale of measurement. However, translating patterns across different scales is challenging because of the scale mismatches of the data. I think this is a good example of how scale can play an important role in GIScience and it is also where the geostatistical approaches come into play.

One doubt that remains to me is related to one of the main critiques of geostatistcal approach(or modelling approach in general), which is the need of prior understanding of the spatial processes. Nowadays as more and more data-driven approaches is available (e.g. machine learning) and challenging the model-driven approach, does geostatistical approach still have a place in data analysis? I would like to learn more about the comparison of the two types of approaches.

I was pleased to read so much about the modifiable areal unit problem (MAUP) in Marceau’s discussion of scale, as last semester this was a constant thorn in my side. In one final project, I had major issues with the aggregate problem he describes. I was trying to correlate homicide locations in Baltimore to different spatial characteristics, namely socioeconomic status, green space, and vacant buildings. I had the location of every homicide in Baltimore in 2017, vacant building in Baltimore, and a layer file for every park and green space in the city. However, for socioeconomic status I only had the average income for large swaths of the city called “community statistical areas;” these were not nearly as granular as the homicide locations, vacant building sites, and green spaces. There were no better spatial data for socioeconomic status that I could find, so these were the data I used. My analysis revealed that, using these data, both vacant building locations and green spaces are significantly correlated to homicide locations but income is not. However, this does not prove that neighborhood income isn’t correlated to homicide incidence as the income data I used are low resolution compared to the other data (see the maps below for comparison). I happened to conduct multiple regression models for this project, the exact type that Marceau cites as being vulnerable to the MAUP: “the authors [of one study on the MAUP] demonstrated that modification of the areal units from which data are collected create a severe problem for parameter estimation in multiple regression models” (1999). Therefore, I was right to be cautious about the conclusions drawn from my own Baltimore analysis. In addition, I will have to keep the MAUP in mind for my GEOG 506 project as my theme is geodemographics and any demographic data I use will certainly be aggregated to some scalar unit.

Scale is a complex topic with numerous definitions encompassed within diverse conceptual frameworks. For example, in human geography, scale may be perceived as a consequence of social behavior at different levels such as household, neighborhood, state, and nation. In this article, the spatial data analysis perspective is at the forefront, and a definition of scale which relates to spatial extent is used through this article. Atkinson and Tate (2000) divide spacial scale into two elements: 1. scale of spatial measurement, and 2. scale of spatial variation. Any analysis of spatial data is dependent on the support (e.g., geometrical size, shape, and orientation of the measurement units) and coverage of the data. Thus, characterizing the spatial scale of variation and how this relates to the measurement scale should be a fundamental part of any application of such data. Overall, this article provides a comprehensive account of spatial scale problems in geographical information science contexts and presents a variety of geostatistical approaches useful for characterizing scales of spatial variation and re-scaling data.

As a transportation geographer, I am particularly interested in the modifiable areal unit problem (MAUP), which lies at the heart of any analysis of spatial aggregation. Although spatial data are increasingly disaggregated, many studies on transportation geography require some level of aggregation (e.g., traffic analysis zones, census tracts, dissemination areas). Atkinson and Tate (2000) argue that MAUP can contribute to a significant loss of information in the aggregation of data in large units of geography. Indeed, the aggregation of the same data in different configurations of spatial units can lead to dramatically different results. Besides of the MAUP, we must also be cautious when we transfer relationship at the individual level to aggregates of individuals. For example, a survey can reveal that men ride bikes more frequently than women. However, one can measure lower cycling rate at census tracts with a higher share of male since these census tracts lack access to cycling facilities.

After reading through D.J. Marceau’s “The Scale Issue in the Social and Natural Sciences” paper, I’m left reflecting on some of the more abstract issues outlined at the beginning of the article. The paper does a good job of summing up the state of the issue of scale in academic geography as of 1999. This of course begs the question of what developments may have occurred around this field since then, and what effects that may have had on geospatial applications as GIS tech has become an everyday part of our lives.

I’m especially interested in this issue in relation to navigation apps and online mapping services for consumer use. The majority of the average person in the developed worlds interactions with geospatial technologies operate at two scales – that of a pedestrian and that of a vehicle, commuting a distance between 20 minutes and 2 hours. These scales are very specific, and we spend much of our lives operating within these bounds. I’m curious about how societal perceptions of space and distance have been affected by this pattern, and how our use of navigational aids may have locked us into certain mindsets about the scale of our lives and our communities.

I’m also curious about exploring the concept of “scaling” as a verb. It certainly has little to do with physical distance between things, as evidenced by the terms use in far more abstract conditions than the geographical sciences. A hierarchical worldview is implied by the use of the word scaling, and its application says to me a lot about how the user sees the world. What is the origin of hierarchical frameworks of organizing non-geographic information? Was is it inevitable that scientists structured the world this way? Did geographic hierarchical structures of thinking influence non-geographic conceptualizations of scale, or was it the other way around? The last question may be one of those chicken-and-egg ontological problems without a solution.

Atkinson and Tate provide a thorough overview of geostatistical issues and their solutions in this piece. Although the authors do make a point of simplifying concepts, I still had to read the article a few times to really understand what was happening. I would have greatly benefited from more examples to illustrate their points, or maybe one that was a bit simpler than the one they chose. Despite my difficulties in grasping everything that was said in the article on my first read, I believe this article to be an important starting point for those embarking on research where scale plays a major role. For instance, knowing about the smoothing problem when performing kriging analyses (and that the variance lost through smoothing can be estimated by subtraction) could be the difference between a successful project and an unsuccessful one. 

Some things I can’t help but wonder after reading this piece are:

What has changed in the literature since this was published in November 2000?

Is there now a solution to the covariance problem (ie that it varies unpredictably between the original and kriged data)?

This article made me think more about how scale can relate to my own project for this course, and what issues to avoid in my own research. I’ll be sure to look at more literature concerning the interaction of scale and VGI moving forward. Overall, I found this piece to be a solid overview of geostatistical problems and solutions.

My initial reaction to the subject of geospatial ontologies was incredulity. Both of the readings I was assigned (Smith and Sinha) were well written, but in many ways seemed to be a stretch to me. Geospatial ontologies strike me as an entertaining philosophical game, similar to something like “the trolly problem.” – interesting to think about and discuss but more often than not irrelevant except in especially extreme circumstances. The subject of ontologies is applicable at a highly abstract scale, and often doesn’t have much to do with practical, day to day geography. I was unconvinced.

This view was challenged during the class on geospatial ontologies. As the breadth of the subject was explored, I began to consider its implications on my own thesis. Many of the models and systems I’m considering studying are built on abstract assumptions of universal ontologies, and require a certain academic geographical fluency to piece together. Ontologies underly all science, including geography, and its important to take them into consideration as further research is done.

After listening to the lecture on geospatial ontologies, readings the literature, and discussing the subject with my classmates, I have developed a slightly broader perspective. However, I have to admit that I do find the subject irritating nonetheless. Geospatial ontologies deal with fundamental discrepancies in the world, and are not easily resolved – meaning that even when they are not the subject of something they underlie the theory supporting it. The fact that its seemingly impossible to define mountain in a clean, universal way throws much of (what has traditionally been) the field of geography into doubt. This in turn makes it hard to feel confident in the outputs of any research in this field, and leaves one dissatisfied and frustrated if you expect a nice logical solution to this problem.

In this article, the authors argue that conventional geospatial ontology in GIScience often seeks to reach universality and interoperability, which could potentially lead to assimilation of indigenous culture. They proposed that some more inclusive and participatory approaches, such as hermeneutics and heuristics, should be applied when developing geospatial ontologies.

While I strongly agree with the authors’ opinion that the indigenous knowledge should have a place in ontologies, it is still unclear to me, how and how much indigenous knowledge we should engage when developing ontologies?  Do we engage all of the indigenous ontology or only a part of them? If we include all of them, would we still be able to achieve the interoperability in ontologies? For example, the place-based multinaturalist approach proposed in the article seems to support the idea that we should not leave out any indigenous culture and there is no universality, but to me, on the other hand, this means that we can hardly achieve interoperability. On the contrary, if we include only a part of them, who gets to decide what to include? Would it be the scientists or the indigenous people?

In my perspective, the development of ontologies or the seek for interoperability per se is more or less assimilation of indigenous knowledge, regardless of what approach we used. The approaches proposed by the article might contribute to less/slower assimilation of indigenous culture, but as long as the development of ontologies is led by the majority (which I assume most of the time it is?), indigenous culture would always be underrepresented. I couldn’t see any way that interoperability and the full engagement of indigenous culture in ontologies could me mutually achieved.