assignment due Thursday, January 28th (please post your comments before 9am so we can read them before class)
The goal of this assignment is to give you an idea of what is going on in “Visualization Research” as a Computer Science Discipline. This is only one perspective on visualization, and this will give you a particular slice of it, but its better than nothing.
The premiere academic venue for Visualization (as a computer science sub-area) is IEEE “VisWeek.” Its a set of 3 conferences that are co-located. The proceedings are published as a special issue of IEEE Transactions on Visualization and Computer Graphics. The event is evolving. Its usually in October.
The past few years, there’s been 3 events “Vis,” “InfoVis,” and “Vast” (Visual Analytics Science and Technology). The most recent one (2009) was in Atlantic City this past October.
The goal of this assignment is to give you an idea of what kinds of things go on at this venue (as a way of sampling what “Visualization Research” is.
Your task is to look through the “proceedings” of the “conference” (really the 3 co-located events) and see what catches your eye. Of course, this being the modern era, you won’t actually look at the printed proceedings (they don’t even give it out at the conference – they give out a USB stick). One down side is that printed proceedings are great to flip through for this kind of purpose, and online proceedings are less skimmable. You don’t need to read the papers, but I want you to get a sense of what kinds of topics are there (and might be interesting to you). If you had the printed proceedings, you could flip through and see what pictures stood out.
What you should do (the resources for doing this are below):
- Look over all the titles, see what catches your eye.
- For some subset of those, look a little more closely. Read the abstract, look at the pictures, maybe the author has a website or something…
- Pick a few of your favorites. Between 3-5. At least one must come from InfoVis, at least one must come from Vis. Give your list as a comment to this message. Please either remember your list or bring it to class.
Without the printed proceedings, your resources for doing it:
- The VGTC website. VGTC is the committee of IEEE that organizes the conferences. They have a great website. For example, at this page you can see a list of all the papers, links to the abstracts, and links to the slides from most of the talks from Vis09. (there’s a similar page for infovis).
- The graphics papers on the web resource has links for Vis09 and InfoVis09. These are unofficial, but they usually have links to either the author’s web pages or the project web pages, where you can find more info (and even the PDF of the paper).
- The official digital library page. Most useful to get the actual papers. We have a campus-wide subscription (so either access it with a campus IP address, or use the library’s proxy server).
- If you’re off campus, you need to access the IEEE DL via a proxy. I think this works.
All you need to do is add a list of 3-5 papers as a comment to this posting, and come prepared to talk about what you’ve found. Again, I don’t expect you to actually read any complete papers (you are welcome to, but there will be plenty of time for that later in the semester) – but I do want you to get a sense of the range of topics that people are writing about.
Addendum: the digital library, while inconvenient, is the only real way to get the papers reliably and officially. Many authors put copies of the papers on their personal or group websites, but not everyone (and its unclear with the IEEE copyright agreement if this is a legal thing to do. it is OK with ACM).
Addendum 2: I understand that the papers people pick will be biased towards those that are convenient to find. There is no notion that this will be an unbiased sampling.
{ 20 comments }
Actually: you can follow links through the proxy server even while you’re on-campus. And you can try sending a URL through the proxy server by appending “ezproxy.library.wisc.edu” to the hostname.
So the digital library page’s URL, proxied, becomes:
http://ieeexplore.ieee.org.ezproxy.library.wisc.edu/xpl/tocresult.jsp?isYear=2009&isnumber=5290686&Submit32=View+Contents
General Notes
————-
a. I found InfoVis to be more interesting than Vis.
b. Since not even the abstracts were easily accessible for most of the papers in Vis, I ignored everything past Session 4.
c. The word “novel” is used by most authors to describe their own presentation. This must be a fairly overused term in the vis community, or perhaps in the computing community in general.
d. I understand that every discipline has its own jargon, fully understood only by its own tribe, but this community seems to be particularly jargon prone. Consider —
“Capitalizing on recent advances in matrix approximation and decomposition, ( Exemplar-based Visualization) presents a probabilistic multidimensional projection model in the low-rank text subspace with a sound objective function.”
Good lord! Now, my english is pretty good by any standard, but I have no idea what the authors are trying to say there.
InfoVis 2009
————
1. Protovis: A Graphical Toolkit for Visualization by Bostock and Heer.
Intrigued by this paper because am familiar with Heer’s work, particularly his Vizster application (I believe that was his Ph.D. thesis). Downloaded the Protovis toolkit, but in spite of the logic of the authors on why such a toolkit was needed (declarative vs. imperative programming) in spite of many similar alternatives already out in the wild, I found the tool to be less than compelling.
2. Spatiotemporal Analysis of Sensor Logs using Growth Ring Maps by Bak, Mansmann, Janetzko and Keim.
Tackles a problem familiar to me in my mapping context, that is, how to symbolize overlapping markers (think residents in a multistory condo). I have been contemplating creating a prototype that disaggregates overlapping markers in an animation, much like Google Earth does with overlapping markers, but in a web-based context. Might be worthwhile as a project for this class.
3. Harnessing the Web Information Ecosystem with Wiki-based Visualization Dashboards by McKeon.
I once tried to create a dashboard for a project, but did not quite succeed. Since dashboards, but their nature, provide a controlled view of some complicated activity with many variables and views, it is hard to imagine their creation to be wikified. Hence, I found this paper to be of interest. It is based on ManyEyes, so it is easy to experiment with. I like stuff I can dig into, get my hands dirty. This paper took me there.
Vis 2009
——–
4. Time and Streak Surfaces for Flow Visualization in Large Time-Var ying Data Sets by Krishnan, Garth and Joy.
Showing motion in a static visualization is a common challenge. The authors present a method for computing such visualizations that they term as “time and streak surfaces.”
gah! Beware, comments can be edited/updated. Once you hit [Submit], its immutable.
where by “comments can be edited” I mean “comments can *not* be edited.”
1. Exploring the Millennium Run – Scalable Rendering of Large-Scale Cosmological Datasets (Vis 2009)
2. Hue-Preserving Color Blending (Vis 2009)
3. Visualizing Social Photos on a Hasse Diagram for Eliciting Relations and Indexing New Photos (InfoVis 2009)
4. Flow Mapping and Multivariate Visualization of Large Spatial Interaction Data (InfoVis 2009)
5. Combining automated analysis and visualization techniques for effective exploration of high-dimensional data (VAST 2009)
1. Focus+Context Route Zooming and Information Overlay in 3D Urban Environments, Vis 2009.
2. Visual Human+Machine Learning, Vis 2009.
3. Hue-Preserving Color Blending, Vis 2009
4. Protovis: A Graphical Toolkit for Visualization, InfoVis 2009.
5. A Nested Model for Visualization Design and Validation, InfoVis 2009.
1. VisMashup: Streamlining the Creation of Custom Visualization Applications
Emanuele Santos, Lauro Lins, James P. Ahrens, Juliana Freire, Cláudio T. Silva
There wasn’t much information available (i.e. none) but this interested me because so much network visualization software (my major field of interest) is custom, takes considerable time to build, and doesn’t necessarily involve best practices in visualization.
2. Interaction Techniques for Selecting and Manipulating Subgraphs in Network Visualizations, McGuffin, M.J.; Jurisica, I.
Much of my work involves making complex networks clear and useful to laypeople. Subnetworks are central to this process, and finding and visualizing subnetworks clearly is an important aspect of this work.
3. “Search, Show Context, Expand on Demand”: Supporting Large Graph Exploration with Degree-of-Interest, van Ham, F.; Perer, A.
This is a very useful paper for the inverse of (2) above. For a variety of reasons, I work with whole community networks (community in this context means literal cities). I want to find ways for people to look at the entire community and see clusters or subnets that are useful to them. This involves multiple problems: what *is* useful to laypeople in exploring complex networks; how do you make this useful material stand out and visible; and how does this vary across users classes?
4. ResultMaps: Visualization for Search Interfaces, Clarkson, E.; Desai, K.; Foley, J.
This appealed to me because my work involves the question of how people find useful networks in communities, and what is the easiest way “in” to this problem: whole network visualization, subnetworks, simple search, or some combination.
1. Loop surgery for volumetric meshes: Reeb graphs reduced to contour trees
2. Exploring the Millennium Run – Scalable Rendering of Large-Scale Cosmological Datasets
3. code swarm: A Design Study in Organic Software Visualization
4. Towards Utilizing GPUs in Information Visualization: A Model and Implementation of Image-Space Operations
5. Visual Analysis of Inter-Process Communication for Large-Scale Parallel Computing
1. Visual Exploration of Nasal Airflow
2. Scalable and Interactive Segmentation and Visualization of Neural Processes in EM Datasets
3. A Nested Model for Visualization Design and Validation
4. Hue-Preserving Color Blending
A few general notes:
* In general, InfoVis looks like more fun than Vis.
* The pages for both sites are surprisingly hard to scan, considering it’s for they’re for vis sites.
And stuff I thought was really neat — or at least, notable:
1: Depth-Dependent Halos: Illustrative Rendering of Dense Line Data
2: ABySS-Explorer : Visualizing Genome Sequence Assemblies
3: Constructing Overview + Detail Dendrogram-Matrix Views
And, harkening back to my library days:
4: Mapping Text with Phrase Nets
I went through the abstracts of the articles I pick. My personal interest is mainly on how to make data telling more. Therefore, I select one with sampling methodology, one with surface descriptor, one with bioinfo background, and one from multidimensional data visualization.
Vis09
1. Sampling and Visualizing Creases with Scale-Space Particles
Gordon L. Kindlmann (University of Chicago), Raál San José Estépar (Harvard Medical School), Stephen M. Smith, Carl-Fredrik Westin (Harvard Medical School)
In improving mesh generation and shape analysis, a good sampling from particle system is important. This paper proposes the role of particle systems in sampling structure from unsegmented data and describes a particle system that computes samplings of features effectively represent many anatomical structures in scanned medical data.
2. Intrinsic Geometric Scale Space by Shape Diffusion
Guangyu Zou, Jing Hua, Zhaoqiang Lai (Wayne State University), Xianfeng Gu (State University of New York at Stony Brook), Ming Dong
Given any signal by observation, features inherently exist at various scales. Intrinsic Geometric Scale Space (IGSS) is a scale space of the Riemannian surface geometry, and it provides a framework for feature extraction and matching with inherent scale information.
InfoVis09
3. Constructing Overview + Detail Dendrogram-Matrix Views
Jin Chen, Alan M. MacEachren, Donna J. Peuquet (GeoVISTA Center, The Pennsylvania State University)
The research proposes a strategy that links an overview dendrogram and a detail-view dendrogram, each integrated with a re-orderable matrix. This tree plots strategy could be useful in many research fields.
4. Scattering Points in Parallel Coordinates
Xiaoru Yuan, Peihong Guo, He Xiao (Peking University), Hong Zhou, Huamin Qu (The Hong Kong University of Science and Technology)
I am particularly interested in that they present a novel parallel coordinates design integrated with points, by taking advantage of both parallel coordinates and scatterplots.
I guess its only fair if i do this too. Some notes I made while skimming the proceedings (i had a physical copy) are at http://www.cs.wisc.edu/graphics/Wiki/Gleicher/Vis09
Me doing this is a little different, since I have different information (I was at the conference, saw lots of presentations and the fast-forward), and several competing interests (find things for class, find things for my own work, things that just seem interesting)…
I too found the infovis papers more interested in terms of their concepts, but also like some of the more mathematical issues addressed in the vis community.
But some of my “favorites” – OK, I picked too many.
* Nested Model for Validation… – good food for thought, we’ll read this
* Bubblesets – just a cute idea, and it looks nice – but i have some doubts about how they did it
* Qualitiative Texton Sequences – good follow-through on a foundational thing
* Route Zooming – I like lots of things about this (problem, approach, details, …)
* Interactive Vis of Molecular Surface Dynamics – I saw the talk, and I still don’t think its possible 🙂 but i’m really interesting in the problem.
* Structuring Feature Space: A Non-Parametric Method for Volumetric Transfer Function Generation – seems like a good idea for a basic problem. pulls out some appropriately sophisticated mathematics.
I read the following five paper which are interesting and have many applications.
Paper 1: Vis2009
Title : Loop Surgery for Volumetric Meshes: Reeb Graphs Reduced to Contour Trees
Authors : Julien Tierny, Attila Gyulassy, Eddie Simon, Valerio Pascucci
Comments: In this paper, authors propose a new technique to construct Reebs graph
using “Surgery Technique” and claim that it is 6500 times faster than
previous known techniques. Which is true remarkable, but the following
sentence “In this case, our technique produces results in matter of seconds
even for the largest models” is rather overstatement.
Paper 2 : Vis2009
Title : Applying Manifold Learning to Plotting Approximate Contour Trees
Authors : Shigeo Takahashi, Member, IEEE, Issei Fujishiro, Member, IEEE,
and Masato Okada
Comments: A Contour tree ( or Sketon tree ) encapsulates evolution of scalar
field which is used to understand topological properties of a field. In
this paper, authors have used a well known nonlinear data
reduction technique “Manifold learning” to construct contour trees
in higher dimension. They also use hierarchical representation
for handling large scale dataset.
Paper 3 : Vis2009
Title : Multi-Scale Surface Descriptors
Authors : Gregory Cipriano, George N. Phillips Jr., Michael Gleicher
Comments : In this paper, authors make an attempt to characterize surfaces
based on curvature based descriptors. They show use of local
shape descriptors in many applications such as shape
matching, stylized rendering, visualization and analysis
of scientific data.
Paper 4 : InfoVis2009
Title : Search, Show Context, Expand on Demand: Supporting Large Graph Exploration with
Degree-of-Interest
Authors : Frank van Ham and Adam Perer
Comments : Exploring information from large sparse graph is challenging. In
this paper, authors believes that by using “degree of interest”,
the complexity of graph exploration could be reduced. They have
developed a prototype on legal documents and showed that it was useful
in extracting relevant documents.
paper 5 : VAST2009
Title : Combining automated analysis and visualization techniques for
effective exploration of high-dimensional data
Authors : Andrada Tatu, Georgia Albuquerque, Martin Eisemann,
Jörn Schneidewind, Holger Theisel, Marcus Magnork, Daniel Keim
Comment : The projection of higher-dimensional dataset to lower
dimensional space is although quite attractive, but its complexity is
too high to rely on manual selection of lower-dimensional views. In
this paper, the authors automate ranking of scatterplots and
parallel coordinate views which has high probability of providing useful
information.
* Depth-Dependent Halos: Illustrative Rendering of Dense Line Data – Vis09
* Perception-Based Transparency Optimization for Direct Volume Rendering – Vis09
* Bubble Sets: Revealing Set Relations with Isocontours over Existing Visualizations – infoVis09
* A Nested Model for Visualization Design and Validation
* Smooth Graphs for Visual Exploration of Higher-Order State Transitions – InfoVi09
1. Perception-Based Transparency Optimization for Direct Volume Rendering
I like to stare at pictures with visible structures.
2. Document Cards: A Top Trumps Visualization for Documents
My friend and I thought that a summary diagram of a paper will be more efficient than an abstract or even the whole paper in the aspect of understanding and time.
3. A Comparison of User-Generated and Automatic Graph Layouts
It is interesting the layouts look so different to each other.
Here are a few papers/presentations that look interesting to me:
InfoVis
Systems
A Multi-Threading Architecture to Support Interactive Visual Exploration
Harald Piringer, Christian Tominski (University of Rostock), Philipp Muigg, Wolfgang Berger
comment: I really appreciate it when an effort is made to extend the technical limitations associated with complex visual projects.
Vis
Perception-Guided Visualization
Best Paper Award: Depth-Dependent Halos: Illustrative Rendering of Dense Line Data
Maarten H. Everts, Henk Bekker, Jos B.T.M. Roerdink, Tobias Isenberg (University of Groningen)
comment: The density of each grouping of line data is really easy digest in a short amount of time.
InfoVis
Text Visualization
Best Paper Award: Mapping Text with Phrase Nets
Frank van Ham, Martin Wattenberg, Fernanda B. Viégas (Visual Communication Lab, IBM Watson Research Center)
comment: Just like with a tag-cloud, the viewer can quickly get a sense of the scope and theme of the topic.
InfoVis
Collaborative Visualization
SpicyNodes: Radial Layout Authoring for the General Public
Michael Douma, Grzegorz Ligierko, Ovidiu Ancuta, Pavel Gritsai, Sean Liu (Human-Computer Interaction and Design, University of Washington)
comment: I thought the node based layout was well executed, although the transitions could be more fluid.
1. On the Visualization of Social and other Scale-Free Networks (InfoVis 2008. Yuntao Jia et al.) : social network analysis is all about finding meaningful patterns, which depends largely on effective condensing(clustering, filtering etc) of the edges.
2. Narratives: A Visualization to Track Narrative Events as they Develop (VAST 2008. Danyel Fisher et al.) : “Visualizing news stories in their historical and social context… by understanding how the major topics associated with them have changed over time”. Just the right topic anybody in my department should look at.
3. VisComplete: Automating Suggestions for Visualization Pipelines (Vis 2008. David Koop) : Pure heuristics, effectively visualized.
4. Configuring Hierarchical Layouts to Address Research Questions (InfoVis 2009, Aidan Slingsby et al.) : Explains how to use hierarchial displays to show multiple aspects of large multivariate datasets. Provides insights to build visual narratives in complex data exploration.
InfoVis
——–
1. Viz-A-Vis: Toward Visualizing Video through Computer Vision
2. Visual Analysis of Inter-Process Communication for Large-Scale Parallel Computing
3. Distributed Cognition as a Theoretical Framework for Information Visualization
4. Vispedia: Interactive Visual Exploration of Wikipedia Data via Search-Based Integration
Vis
—-
1. BrainGazer – Visual Queries for Neurobiology Research
– “Hue-Preserving Color Blending”
I don’t have any experience at all with this problem, yet the main idea of the paper is very simple, and clea
rly
described. The results were convincing too.
– “Interaction Techniques for Selecting and Manipulating Subgraphs in Network Visualizations”
I like this one because graph data is ubiquitous, and though I couldn’t find the paper, the demo looked like
it allows
some useful interactions.
– “Perception-Based Transparency Optimization for Direct Volume Rendering”
This one is interesting because it is an interesting problem I have never considered, and it may be useful fo
r
visualizing brain scan data. Fig. 8 shows their approach on a head CT scan.
– “Applying Manifold Learning to Plotting Approximate Contour Trees”
This one is interesting just because it is close to my research area.
-Molecular Surface Abstraction (Vis 2007)
Greg Cipriano, Michael Gleicher (University of Wisconsin, Madison)
– Mapping Text with Phrase Nets
Frank van Ham, Martin Wattenberg, Fernanda B. Viégas (infoVis 2009)
-Multi-Scale Surface Descriptors
Gregory Cipriano, George N. Phillips Jr., Michael Gleicher (Vis 2009)
-A User Study to Compare Four Uncertainty Visualization Methods for 1D and 2D Datasets
Jibonananda Sanyal, Song Zhang, Gargi Bhattacharya, Phil Amburn, Robert J. Moorhead (Vis2009)