Geospatial Visual Analytics @ GIScience 2008 - Discussion topics

Session 1

paper 1.1

  1. What representation techniques are suitable for the display of the attributes of archaeological sites and what are their advantages and disadvantages?
  2. How effective is the extended STC environment for Archaeological data?

paper 1.2

  1. The visual representation of temporal data
  2. How to display multivariate spatiotemporal data (value patterns vs. spatial patterns)

paper 1.3

  1. Magnitudes and vectors from classical mechanics/dynamics---such as velocity, acceleration, momentum, and curvature---can be calculated as additional, derived attributes of N-dimensional point records in a time-ordered data set. Which of these metrics might facilitate specific visual exploration and analysis tasks?
  2. How can visual analysis tools encode, in different views/ coordinate spaces, both points and (sub)sequences as a function of individual metrics or combinations of them? How does simultaneous encoding of multiple different but related metrics (possibly including location and time) in a limited number of graphical channels affect perception and interaction?
  3. How should visual analysis tools deal with (additional) absence/ noise/uncertainty in the form of invalid values, zeroes, singularities, etc. in the calculated sequence of values for such metrics?

paper 1.4

  1. In order to aggregate and summarize movement data, it is often (always?) necessary to coarsen the spatial component of the data, i.e. use areas instead of points. How (according to what criteria) can such areas be appropriately defined? Two cases should be considered: (a) areas covering the whole territory; (b) areas representing "significant places".
  2. Is it possible to aggregate and summarize massive movement data without dividing the space into compartments or defining "significant places"? How this could be done?

paper 1.5

  1. Summarization algorithms used for space, time and space-time typically rely on (arbitrary?) threshold settings. How do we validated found solutions based on "empircially based" threshold settings?
  2. How can algorithm settings and/or thresholds that are individually valid for space or time problems be combined for space-time problems?

Session 2

paper 2.2

  1. Permitting temporal lags of less than one year would, for this analysis, result in several more matrices and maps. Questions including within year variability could be addressed, yet computation and interpretation investments would increase. In data mining, we don't necessarily know the appropriate temporal aggregation unit (lag) for analysis. Similar to our choice of a global statistic, Ripley's K, to guide the spatial lag parameter of local spatial statistic calculation, how might we visually or computationally introduce a choice of temporal lag parameter for attribute change calculation in the tool?
  2. Immigrant destinations are not emphasized in the user interface of the tool, but rather presented as a drop-down menu. Not emphasizing the spatial arrangement of destinations permits the tool to be directly applicable to spatio-temporal phenomenon that do not have an origin-destination component. However, does this design present a significant limitation for data mining patterns of immigration?

paper 2.3


paper 2.4

  1. Graphical conventions appear to us every day and everywhere. We learned them to the point that they become a natural language. For example, green, yellow and red, immediately remind us of the traffic lights indicating a level of danger. Because those signs become so natural in our lives, would it be more efficient to use such graphical conventions (even if they don’t correspond to semiology theory) for information to pop up than good semiology?
  2. To what extend can we customize a visual representation so that it lowers the attention towards details and puts forward the trends in the data without any mathematic or statistic manipulations? In other words, the goal is to direct the user’s attention to the overall reading level allowing, even forcing, visual aggregation of data and enlightenment of information simply by changing visual displays of data. A good example would be to visually group data by displaying all middle categories the same way and displaying only the extremes categories differently. This would places the emphasis on extremes and might help to discover some patterns without changing the number of categories.
  3. “Business graphics are more colourful, easier to produce and faster than ever, but the popular basic chart types haven’t changed in decades. Numerous new visualization techniques have been introduced , but none have been widely adopted, despite the apparent need for them in business intelligence applications with large data volumes to report. One problem seems to be that the more advanced charts are not immediately intuitive for the reader; if users need special training to understand new chart types, they are not likely to become mainstream.” (Business Application Research Center, 2007) Along the same line, are new or unusual geovisualization techniques “really” useful for non-expert users or decision makers?

paper 2.5

  1. How can you aid a user in associating geographic context with annotations made in a geovisual analytic environment to better support capturing and sharing insights derived through the use of the application?
  2. How might annotations support collaborative use of a geovisual analytic tool and what tasks can visual representations of this content help the user perform?
  3. How can computational approaches aid the user in making sense of annotation repositories as they grow over time and augment the use a geovisual analytic environment?

Session 3

paper 3.2

  1. when interpret the clusters generated from statistics methods, what other issues we need to consider, other than reliability and heterogeneous issues;
  2. how to judge ( or evaluate) the success of the methods that combine visualization and (spatial) statistics methods

paper 3.3

  1. Clustering is often accompanied with the problem of visually defining size and boundaries associated with the clusters. What are some of these problems, and how do they influence how researchers and the general public view clusters of events?
  2. What other types of methodologies are available for visualizing clusters in meaningful and statistically significant ways?

paper 3.4

  1. How do geovisual analytics yield more effective hurricane climate analysis than traditional weather science approaches?
  2. What are the current limitations of the system described in this presentation and what potential solutions are envisioned?
  3. How does the system facilitate space and time analysis?

paper 3.5

  1. Our VA infrastructure provides the opportunity to collaboratively integrate data and efficiently stream the differences or past revisions of this data.
    • What are the spatio-temporal challenges we should deal with? E.g., we see possibilities with the aggregation of time intervals or interval steps to compare data.
    • What interfaces or methods should we support to best assist GVA and its related data mining, statistics, and optimisation tasks?
  2. Imagine globally spread research teams performing collaborative GVA.
    • What kind of data would they dynamically add to the shared repository? E.g., one might upload a large set of statistical data, one might tag a selected area with some text notes, and one might add a figure or bitmap data;
    • What is the degree of interactivity? Is it a asynchronous process which could be compared to some kind of GVA Wiki(pedia)? Or is it a synchronous process more comparable to a chat room than a Wiki;
    • Are there established workflows with different roles for collaborative GVA (as seen with other collaborative document authoring tools)?

Session 4

paper 4.1

  1. The need for quantitative evaluations of geovisualisation designs in order to yield valid scientific results.
  2. Studying “internal representations”, neurocognitive principles, and cognitive processes with respect to the design and understanding of geovisualisation displays as well to the development of geospatial visual analytics tools.
  3. The effect of information density (number of items displayed) on the user. Is 'less more' or is 'more more'?

paper 4.3

  1. How to best enable geographic collaboration for visual analytics, in terms of a user interface?
  2. Which visual analytic operations should be exposed for sharing between collaborators, and which are best reserved for a single analyst?

paper 4.4

  1. The effectiveness and usability of the structured phrase-driven approach for data query and visualization for non-IT users.
  2. What are primitive representation techniques for geo-spatial visualization in terms of properties and objects in oil reservoirs?