TMPA-2017: Layered Layouts for Software Systems Visualization

Layered Layouts for Software Systems Visualization
using Nested Petri Nets
Alexey Mitsyuk Yaroslav Kotylev
National Research University Higher School of Economics
Faculty of Computer Science
Laboratory of Process-Aware Information Systems (PAIS Lab)
amitsyuk@hse.ru, yavkotylev@edu.hse.ru
Dolgoprudny
March 4, 2017
Alexey Mitsyuk, Yaroslav Kotylev (HSE) Layered Layouts for Visualization of NPNs Dolgoprudny March 4, 2017 1 / 26

Overview
1 Introduction
Software Visualization
Layered Layouts
Similar & Related Approaches
2 Visualization of Nested Petri Nets
Basic Concepts
Overview of Our Approach
Algorithms
Concluding Information
3 Visualization of Petri Nets with Data
4 Conclusion

Introduction Software Visualization
Software Visualization is Useful (Captain Obvious:)
Motivation
Typical modern software has complex architecture
Visual analysis can be insightful
Our problem
How to make visualization comprehensive and readable at the same time?

Introduction Layered Layouts
Our solution: Layered Layouts (2.5D)
Diﬀerent dimensions (multiple perspectives / ﬂows) in a single view
Intuitive hierarchy visualization
Existing 2D layouts for layers
May be extended with dynamics and animation
Relatively easy to implement

Introduction Similar & Related Approaches
Similar Approaches - 1
(all images are the property of their authors)

Similar Approaches - Summary
1. Ideas similar to layered layouts (2.5D) are known in graph drawing.
2. 3D methods were already applied in software visualization.
3. Though the idea is not quite new, it reincarnates regularly in different
fields.
4. There are o lot of works on visualization of program structure (classes /
packages) and program executions (traces). We visualize flow models.
5. In the remainder we follow this idea and propose the 2.5D layouts
approach for nested Petri nets visualization.

Visualization of Nested Petri Nets Basic Concepts
Visualization of Nested Petri Nets (1)
Basic Petri Net (P/T-net)
A PN is a bipartite graph.
Elements: places, transitions, arcs, tokens.
A marking (distribution of tokens) determines the current state.
Transitions denote the action elements. Places denote conditions.
Nested Petri Nets [Lomazova, 2001]
NPNs are a special type of high-level (coloured) Petri nets.
1. A NPN consists of a system net and element nets.
2. Some tokens of a system can be the element nets (nets within nets).
3. In general, an element net can be a system net itself.

Visualization of Nested Petri Nets Overview of Our Approach
Our approach
1. Each net can be shown in a separate layer.
2. Lines between layers connect element net representations with
corresponding tokens.
3. The existing layout is used for layers.
4. We use colouring to improve readability (it is optional).

Visualization of Nested Petri Nets Overview of Our Approach
Interactivity
1. Each layer can be shown or hidden.
2. Hierarchy can be visualized easily.
3. We use zoom in and out functions.
4. We use colouring to improve readability (it is optional).
5. It is possible to rotate/move the model.

Visualization of Nested Petri Nets Algorithms
Layered Layouts Algorithm
Algorithm 1: Layered layout algorithm
Data: Net = (SN, EN) — NPN, where SN is a system net, EN is a set of
element nets
Result: All nodes are arranged
1 setCoordinates(SN,0);
2 layerShift ← computeLayerShift(Net);
3 foreach N in EN do
4 setCoordinates(N,layerShift);
5 end
6 return coords
— The order of foreach traverse can signiﬁcantly change the visualization.
— Fixed layerShift can be calculated based on the size and conﬁguration
of the model (possible improvement — variable layerShift).

Single Layer Layout
The setCoordinates() calls the method, which implements the
Sugiyama-style algorithm (PNs are directed bipartite graphs).
The particular algorithm has been proposed in [Nikitina Mitsyuk, 2015].
Algorithm 2: Petri net layout algorithm
Data: Nodes — List of all nodes
Result: All nodes are arranged
1 Find all connected components;
2 foreach connected component Comp do
3 CNodes ← GetAllNodes(Comp,Nodes);
4 ICNodes ← SearchForInitialNodes(CNodes);
5 SetColumnForInitialNodes(ICNodes);
6 SetColumnForEachNode(CNodes ICNodes);
7 coords ← SetCoordinatesForEachNode(CNodes);
8 end
9 return coords

Traversal Order (based on BFS)

Visualization of Nested Petri Nets Concluding Information
Revealing the Hierarchy
a)
b)
Figure: Manually prepared plane picture a) and an automatic layered visualization
in 3D b).

Visualization of Nested Petri Nets Concluding Information
Prototype Tool
Prototype for NPNs
The prototype has been implemented using Processing [Reas Fry, 2014].
We use Carassius [Nikitina Mitsyuk, 2015] for Petri net editing.
Want to try?
The code with example nested Petri net is acceptable at
https://github.com/yavkotylev/NestedPetriNetsTool
Video demo
Demo link (internet connection required)

Visualization of Petri Nets with Data
Another Case: Visualization of DPNs
Petri net with Data (with K.Samsonova)
Petri nets with data (DPN) were proposed in
[de Leoni and van der Aalst, 2013]
— Only 2 flows: control- and data-
— Control-flow is a P/T-net
— Sparse graphs for data-flow in most cases
— Dynamic colouring (firing, read and write operations)

Conclusion
Future Work (To Do:)
1. To propose the approaches for cases with complex links between layers
2. To introduce a variable shift between layers
3. To evaluate diﬀerent orders for layer drawing
4. To evaluate the force-based algorithms for 2D layers in 2.5D setting
5. To apply layered layouts for diﬀerent modelling notations (e.g. UML
Sequence diagrams, BPMN 2.0)
6. To construct an algorithm for automated camera angle selection &
rotation
7. To add dynamics (model runs)

Conclusion
Conclusion
In this talk we
1 recalled the usefulness of informal visual analysis;
2 showed the idea for visualization of graph-based software models;
3 discussed related and similar approaches;
4 demonstrated how layered layouts can be applied to NPN models;
5 outlined directions for future work.

Conclusion
References
The full list of references can be found in the paper.
M. de Leoni and W. M. P. van der Aalst (2013)
Data-Aware Process Mining: Discovering Decisions in Processes Using Alignments
SAC 2013 Proceedings, pages 1454 – 1461. ACM, 2013.
Irina A. Lomazova (2001)
Nested Petri nets: Multi-level and recursive systems
Fundam. Inform., 47(3-4), pages 283 – 293.
N. Nikitina and A. Mitsyuk (2015)
Carassius: A simple process model editor
Proceedings of ISP RAS, 27(3), pages 219 – 236.
Casey Reas and Ben Fry (2014)
Processing: A Programming Handbook for Visual Designers and Artists
The MIT Press, 2014.

Conclusion
Thank you!
https://github.com/yavkotylev/NestedPetriNetsTool

TMPA-2017: Layered Layouts for Software Systems Visualization

More Related Content

Viewers also liked (20)

Similar to TMPA-2017: Layered Layouts for Software Systems Visualization (6)

More from Iosif Itkin (20)

Recently uploaded (20)

TMPA-2017: Layered Layouts for Software Systems Visualization