Position Paper by Vladimir Batagelj, University of Ljubljana

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?

Such a general infrastructure should be

portable - running on all main platforms;
open - source code and documentation available to the users;
expandable - easy to add new features;
efficient - includes efficient algorithms;
scalable - can be used on small and (very) large datasets;
cooperative - easy to combine with other tools, support for main non-native data formats;
CGI - can be used as CGI program on the web;
user friendly.

Visualization tools are tightly connected with the corresponding analytical tools. Where to put a border? In Pajek, for example, we decided to provide the connections to R and SPSS for statistical analyses of data derived from networks. Also, we are not supporting in Pajek the high complexity algorithms applicable only to (very) small networks, but we provide file exchange with UCINET - a program that contains all these procedures.

Another remark is that most efforts in visualization are still 'paper' (letter, A4) oriented. The general infrastructure should provide also 'movie' outputs and 'vislet's - interactive layouts/pictures.

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)?

Because of great diversity of potential users (needs, IT competencies, ...) it seems that the 'MS Office'-like approach would provide the right answer:

library of very efficient algorithms that can be used through an API in different programming languages;
programming language above the library (some existing language can be used/extended: Python, Javascript, R, ...);
GUI interface(s) to the library for non-programmers and special user-groups (for example, genealogists).

The experiences with the development of statistical environment R show that a scripting language can activate a number of users to contribute (packages) to the development of the system - thus becoming an integration platform for the field.

Very important are also publicly available, well documented datasets (in standard format - VisML ?).

Part II

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

II.1) Project Name and Web Address

Pajek - Program for Large Network Analysis
http://vlado.fmf.uni-lj.si/pub/networks/pajek/

II.2) Core Team Members

Vladimir Batagelj, vladimir.batagelj@uni-lj.si
Andrej Mrvar, andrej.mrvar@uni-lj.si
Matjaž Zaveršnik, matjaz.zaversnik@fmf.uni-lj.si

II.3) Project Start Date: November 15, 1996 (see history) based on previous experiences (see PP871 / introduction)

II.4) Targeted User Group: Researchers non-programmers (for example, social scientists)

II.5) Supported User Tasks: analysis and visualization of large networks, see PP871

II.6) Major Features of the System Architecture: efficient algorithms and data structures connected through the calculator paradigm, see details in PP871

II.7) Algorithms Provided: see PP871 and refrence manual

II.8) Snapshot of the Interface: See here.

II.9) Development Platform: Delphi

II.10) Supported Operating Systems: Windows, runs also on Linux via Wine

II.5) Software Dependencies/Required Libraries: no

II.5) Current License: free for non-profit use

II.5) Number of Users/Downloads: some hundreds, try Google with Pajek network analysis

II.5) Pros and Cons: ask users

II.5) Planned Work: support for line partitions and multirelational networks; support for GraphML; better support for temporal networks; replace macros with scripting language; Linux version (not soon).

Part III

Unfortunatelly I am not able to participate physically in the workshop

Created by Vladimir Batagelj on Thur Oct 7, 2004