Quantitative Evaluation of Hand-Drawn and Automatic Graph Visualizations



Several measures (such as minimizing edge crossing, edge bends, overlaps, etc...) are introduced as drawing principles for network layouts. However, these measures usually conflict to each other and cannot be fully and simultaneously satisfied. This project aims to categorize conventional drawing techniques basing on the scale of graph sizes and find out a composite approach that can be used as a recommendation for the target input data.


In this project, the student has to investigate several conventional graph drawing techniques and does several experimental tests. Basing on the investigation, an algorithm is expected to integrate conventional approaches to draw a graph as nicely as we can draw by hand. The evaluation process should be set up as a web-based application.


  • Knowledge of English language (discussions, source code comments, final report should be in English)
  • Knowledge of web based APIs, like JavaScript, d3, is advantageous
  • Knowledge of C++ and Qt is advantageous
  • Knowledge of Graph Libraries such BGL, GraphViz or Tulip is advantageous
  • Commitment to collaborate with international researchers


The project should be implemented as a web-based application (to be discussed).

This work will be done in collaboration with the Algorithm group in our institute.


For more information please contact Hsiang-Yun Wu (wu@cg.tuwien.ac.at).