Position Paper by Chris North

For the Workshop on "Information Visualization Software Infrastructures" at IEEE 2004 Visualization,
Organized by Katy Börner, Indiana University, USA and Jean-Daniel Fekete, INRIA, France

Part I

I.1) What functionality should a general InfoVis infrastructure provide?

I.2) What do you see as the main technical challenges for creating a central but flexible and universally useful (information) visualization software infrastructure (as opposed to 100 different ones)?

The main problem is the tension between generality and specialization.  "Universal" and "useful" are opposing forces.  Component-based toolkits provide generalized visualization components that can be applied in a broad array of domains and are somewhat useful.  But real visualizations in real data domains invariably require nonstandard features that are very specific to that domain, and not useful or relevant to other domains.  Hence, component toolkits generally become a prototyping tool to explore general visualization design options, and then final visualization designs must be coded from scratch with all the detailed functionality.  These specializations then rarely contribute back to the toolkit because it doesn't fit well into the toolkit's model or requires significant hacks to the model.  This is a general problem is software engineering, but is particularly problematic in visualization due to the inherent creativity, innovation, and variation in visualization design.  Perhaps a potential solution is a framework that covers the full hierarchy of levels of functionality.  For example, high-level components could be plugged together to prototype a design, then a component could be taken apart and customized for the specialized functionality using lower-level constructs.  It is unclear if this solution would work well, or is even feasible.

 

Part II

Please describe the (information) visualization software infrastructure you are working on.

II.1) Project Name and Web Address

Snap-Together Visualization, http://dude.cs.vt.edu/snap/

Fusion is the next generation.

II.2) Core Team Members (Please list in order, Role of Project Member, Full Name, E-mail. eg: Developer, John Doe, jdoe@univ.edu)

PI: Chris North

Development:  lots of students....

II.3) Project Start Date

2001

II.4) Targeted User Group

database owners and users, infovis designers.

II.5) Supported User Tasks

creating customized multiple-view visualizations for relational databases;  discovering data relationships;  publishing data visualizations on the web.

II.6) Major Features of the System Architecture

Coordination model for multiple view visualization based on the relational data model;  diagrammatic visualization schema user interface based on data schemas;  works on any relational database;  can combine multiple databases simultaneously, automatic query execution across databases based on user interactions; component-based approach for plug-n-play visualization by end-users;  simple API (3 functions) to easily integrate components;  3rd parties can contribute components; web-based, runs as applets in web browsers;  created visualizations can be published to the web;  integration of data mining;

II.7) Algorithms Provided

Algorithm to generate join and selection queries across multiple tables and databases based on user interactions;  multi-view coordination algorithm;  Some components include:  parallel coordinates, maps, flexible tree generation from categorical data, ILP data mining.

II.8) Snapshot of the Interface

example using a census demographics database:

II.9) Development Platform

Java, applets.

II.10) Supported Operating Systems

Anything that runs java applets in a browser.

II.5) Software Dependencies/Required Libraries

Java, Swing, ODBC

II.5) Current License

none

II.5) Number of Users/Downloads

Since it runs as applets in a browser, its not really intended for 'download'ing except transparently in the browser.  We have not tracked its usage.  We have given the source to a small number of others for research use.

Our primary test users at this time are our life-science collaborators on a bioinformatics data visualization and mining project.

II.5) Pros and Cons

Pros:  demonstrates the intended concepts:  dynamic end-user plug-n-play environment;  web-based visualization publishing and sharing;  diverse coordinations;  easy to add components with ultra-simple API;  runs in the browser, no need for 'download';  generic to relational databases;  simple visual programming environment.

Cons:  I'm not a fan of Java applets anymore!  Browsers are extremely temperamental with java and applets;  slow;  data access through the ODBC bridge is clumsy and difficult for users to configure;  since components are at the "view" level, the only way to customize lower level features is to implement and add a whole new view component.

II.5) Planned Work

Integration of data mining to further generalize the coordination concept and help users discover relationships;  claims-based visualization design for guiding component selection;  applications in bioinformatics;  moving away from Java applets.

 

Part III

Please describe your main interest in participating in the workshop

Determining the feasibility of combining efforts to create one common, shared IV infrastructure as opposed to 100s of underfunded or proprietary toolkits, platforms and frameworks. Scouring for ideas for a common data protocol for communication between plugins.

Amen!

 

Please use no more than 4 pages, in this HTML format if possible.
Send the completed paper by Sept. 30, 2004 to katy@indiana.edu and Jean-Daniel.Fekete@inria.fr.


Created by Jean-Daniel Fekete and Katy Börner on Thur Aug 12 11:15:27 2004