PROJECT DESCRIPTION

[Title] Local Iterative Optimization for Graph Bundling

[Short abstract] In this thesis, we presented a novel edge bundling technique based on a Physarium approximation of a Steiner tree. This Steiner tree is then used as a routing structure for graph paths. The resulting drawings consist of densely bundled graphs that reduce node-edge overlaps and make it easier to get an overview of the graph data.

[Type of Project] Bachelor thesis

• Project start date: 20.10.2022
• Project due date: 20.02.2022

Supervisors

• Project Owner: Jan Straub
• Supervisor: Filip Sadlo
• Assistant: Maksim Schreck

Status

• Registrated (20.10.2022)
• Work submitted (20.02.2023)
• Final presentation done (13.03.2023)
• Uploaded all materials (13.03.2023)

Installation

This guide only applies to MacOS.

Make sure to have python installed and to download all required packages first, i.e. run:

pip install -r requirements.txt

Usage

WARNING: Change the PATH (below) before running the code. If you just want to test if everyting works correctly navigate to the code folder and run:

python3 main.py

In the folder plots should new be a new file with the default test graph.

Data

In the folder data you will find the different JSON graph files we used.

Create own data

If you want to create your own graph data go in the helper folder. In the create_graph_data.py file you can define the number of nodes, number of edges and the size of the graph. After that duplicate a JSON file and paste the output in the right place.

Transfere data

If you want to test the JSON data on other graph bundling approaches (e.g. Edge-Path Bundling) use the transfer_json.py file to change the format of the data and paste the output in the right place.

Parameters

To generate own plots you have to edit the main.py file. At the end of it you find the all parameters that can be changes.

Path

Change the PATH variable to your path

JSON_FILE_NAME

Change the JSON_FILE_NAME to the name of the JSON data you want to use

PLOT_SELECTION

Choose the mode in which the algorithm should run.

• 0: The algorithm calculates a new graph from the JSON data
• 1: If you already calculated a graph with the selected JSON data and just want to try a different post processing method, you can use this option. The algorithm selects the saved network from the folder savedNetworks
• 2: Allows you to plot the unbundled graph

POST_PROCESSING_SELECTION

Choose the look of the result

• 0: Plot the underlying Steiner tree structure
• 1: Plot the Bezier curve bundling (default)
• 2: Plot the cubic spline bundling

SMOOTHING

Allows you to choose a different smoothing factor. Is unstable for values above 2.

• 0: Steiner points are the control points
• 1: The middle of each edge is the control point

Known issues

• There is an math error issue that sometimes comes up, just rerun the code.