Powered by Highslide JS

My Works

Here is a selection of the projects I was or am involved in.

JFlowMap

JFlowMap is a prototype visualization software which I develop for experimenting with different flow map visualization techniques. A flow map is a representation of the quantities of objects or living beings moving from one geographic location to another (without necessarily showing the real paths which the objects follow). If the number of flows which must be visible at the same time is really big, then visualization becomes cluttered and unreadable. Currently, I am experimenting with techniques like edge bundling and node clustering which help to reduce the cluttering and make the properties of the data visible.

Refugees in 2008. The colors show the directions in which people move: the flow origins are marked red and the destinations are marked green. Flow widths correspond to the numbers of refugees (on a log scale). Click [1:1] to see the full size image.  Migrations between US cities in 2000. The flows on this image were bundled using the force-directed method proposed by D. Holten in 2009. The bundling algorithm runs a physical simulation in which attracting forces between the flows bend them so that they come closer to each other and as a result the visualization becomes more readable. In my prototype I implemented a slighly modified version of the algorithm which takes flow directions into account when bundling. 


CGVis

It is a generic data visualization tool with an innovative zoomable user interface and animations which help the user find particular features in his data. We worked on this project in a team of two developers.

         


Here are two screen records: a smaller one and a bigger one (the animations are notably smoother in the bigger one).

The CGVis binaries and the source code can be downloaded from the project web site.

Adele

I had just about ten days for experimenting with the Tobii Eye Tracker and it's public API, so I made this simple program. In the screenshot you can see a page from Wikipedia with the eye tracking status shown below (both of the eyes, fixation number and the behaviour type). As the user reads the paragraphs they become more and more green. The statistical algorithm which detects the user's behaviour type (scanning, scimming, reading, or learning) from the eye tracking data was developed by my colleague Jürgen Pripfl.

 


Here is a flash screen record (of me reading Wikipedia).

BirdEye

BirdEye is a Flex visualization library which follows the approach to building visualizations described by Leeland Wilkinson in his book "Grammar of Graphics". BirdEye can be seen as a language for creating visualization which not only provides a set of predefined visualization types, but also allows the users to create new kinds of visualizations easily.

My contribution to BirdEye so far has been to make the animated graph layouts from the previous library version work within the new framework based on the Grammar of Graphics.

       


ProClassify

An implementation of the feature selection and classification method which was developed at the Institute for Genomics and Bioinformatics (TU Graz). The method was meant originally for cancer detection on proteomics mass-spectroscopy data, but appears to be useful for completely different datasets too.

ProClassify is free for non-commercial use and can be downloaded from the Project Website.

                   

ProClassifyWeb

A Web version of ProClassify developed to facilitate sharing of datasets and computational capacities by the users. The features of the web version are mostly identical to the standalone version except the data visualization which is missing in the web version.

The online version is available here. And here is the User's Guide.

                   

FHLiteEditor

Screen or presentation manager which allows to create layouts from videos, images, flash movies and other media for a screen and then to schedule them.

My contribution to this project were the visual layout and schedule editors (i.e. the most interesting parts of the GUI :).

                   

WiBi

This was a web application used for collaborative collection and consolidation of study- and research-relevant data by the employees of the university I worked for. After the data is collected the app provided a statistical overview of the "intellectual capital" of the university.

The trickiest thing was to design a flexible model that allowed to configure in the administrative console what kind of data must be collected depending on the employee's occupations. In order to achieve that the adminstrator had to define data-structures using a simple type system in the admin console and also specified who must enter these data. This information was then transformed into UI elements and forms when the users logged in to the system. There was also a hierarchy of data consolidation: after the normal employees entered their data their managers revised it, added more information and then released it for the next hierarchy level. So it was a pretty complex system.

I designed and developed this application on my own. While developing it I concentrated on the domain model and tried to desing it in the simplest and cleanest possible way achieving a good layer separation (e.g. the domain model is completely free of the application/infrastructure level code and there is no business logic in the web and database tiers). After I quit this job I received very good feedback about the code I wrote for the app from my successor who had to maintain it.

                   


Joanna

It's an administration system for students, lecturers and employees of the university I work at. Students are able to evaluate courses, register for exams, see their marks, print certificates etc. Lecturers can enter their courses' details, employees enter their personal data and so on.

             


Watch the screen record showing the course details part of the app.

Wortschatz

A .NET Pocket PC application for learning words and phrases in any foreign language (thanks to Unicode). It uses the classical "learning cards" method: the user sees a word or a phrase and has to guess the translation.

     

phpObjectForms

An open-source object-oriented library for HTML forms processing.

See the phpObjectForms Project Site at SourceForge for details.

Daisy

A web-application framework for eCommerce web sites featured with back-office, inventory, customer and order tracking, credit card processing, numerous reports etc. Here are some screenshots of the back-office:

     

SiteMaker

A small helper tool allows to easily build small static XML-based web sites facilitating XSLT for HTML generation. Several websites are currently built und updated using this tool: I also used this tool to generate help pages for ProClassify and ProClassifyWeb.

SimQA

A prototype meta search engine which I wrote for my thesis to evaluate several algorithms that can be used by a Question Answering system. I then reported the results of these experiments at the RCDL conference (Slides).