PAOHvis: Analyzing Dynamic Hypergraphs with Parallel Aggregated Ordered Hypergraph Visualization

On this page... (hide)

1. Description
2. Publications
3. Videos
4. AVIZ and ILDA Coauthorship Network
5. Online Demo
6. CSV format
7. Case Studies Using PAOHVis
8. Other Works Using PAOHVis
9. Comparison with UpSet
10. Source Code
11. Acknowledgements
12. Contacts

1. Description

Parallel Aggregated Ordered Hypergraph (PAOH) is a novel technique to visualize dynamic hypergraphs. Hypergraphs are a generalization of graphs where edges can connect more than two vertices. Hypergraphs can be used to model co-authorship networks with multiple authors per article, or networks of business partners. A dynamic hypergraph evolves over discrete time slots. A PAOH display represents vertices as parallel horizontal bars and hyperedges as vertical lines that connect two or more vertices. We believe that PAOH is the first technique with a highly readable representation of dynamic hypergraphs without overlaps. It is easy to learn and is well suited for medium size dynamic hypergraph networks such as those commonly generated by digital humanities projects - our driving application domain.

slides

2. Publications

Valdivia, P., Buono, P., Plaisant, C., Dufournaud, N. and Fekete, J-D., Analyzing Dynamic Hypergraphs with Parallel Aggregated Ordered Hypergraph Visualization. IEEE Transactions on Visualization and Computer Graphics, Institute of Electrical and Electronics Engineers, In press, ⟨10.1109/TVCG.2019.2933196⟩.

Valdivia, P., Buono, P., Plaisant, C., Fekete, J-D. and Dufournaud, N., Using Dynamic Hypergraphs to Reveal the Evolution of the Business Network of a 17th Century French Woman Merchant, Proceeding of the VIS4DH Workshop (2018) 1-5

Laurent Beauguitte. Paohvis, un outil pour l'exploration visuelle d'hypergraphes dynamiques. 2020. ⟨halshs-02570627⟩

3. Videos

PAOHvis video

IEEEVis 2020 Fast Forward (30s) https://youtu.be/OvvZxaoKVzU

Virtual presentation at IEEEVis 2020 https://youtu.be/Ilvo6wdi1KA

4. AVIZ and ILDA Coauthorship Network

5. Online Demo

You can play with a new web prototype version of PAOHVis by clicking on this link.

An old version of PAOHVis can be found by clicking on this link.

You can also run it locally if you have a local web server such as python. Download the zip file, unzip it, and start a web server from the `paohvis` directory like that:

python -m http.server

Then, on your browser, open the following URL: http://localhost:8000/paoh.html

6. CSV format

To create your own dynamic hypergraph, use a spreadsheet program or your favorite text editor and save it into a CSV file encoded in UTF-8. Be careful, Microsoft Excel does not generate UTF-8 for international characters, but LibreOffice does. Make sure that the separator character is either a coma (,) or a semi-colon (;) since some spreadsheet programs will insist in using colon, or a tabulation character instead. Even the slightest error in the CSV file will prevent PAOHVis from loading your file.

The CSV file should contain the following columns, with an optional header line. The optional header line must be all in caps, otherwise it will be interpreted as a data row. The first line of your CSV file should contain valid data row or the header.

edge_id, node_name, time_slot, edge_name_description, group_name, role

About the CSV format (assuming a use case of documents and people)

Each row is a link between a document and a person:

1st column: the id of a document (that mentions people). It can be a number or a string.
2nd column: the name of a person
3rd column: the date or really the name of the "time-slot".

Note that if you use periodic time slots (e.g. years or months) some months may be empty and the empty time-slots will not appear in the visualization. If you don’t want that to happen you can create a fictitious person "_" who appears in a fictitious document in those empty months.

4th column [optional]: the name of the document. It is shown when the tooltip option is ON.

If the name of the document associated with the same document id is different between rows, the last one will be shown in the tooltip.

5th column [optional]: the group the person belongs to, for this time slot.

If the group of the person associated is different between time slots, the last one will be shown on the left.

6th column [optional]: the role the person belongs to, for the time slot of this document.
The rest columns are ignored

The three last columns can be left blank for all rows.

Example

This is a csv example with 6 documents, 5 people, 4 time slots and 2 groups:

1,John,2000,Meeting,AAA,Participant
1,Mary,2000,Meeting,B,Organizer
1,Alex,2000,Meeting,B,,Participant
2,Mary,2000,Contract,B,Seller
2,Alex,2000,Contract,B,Buyer
3,John,2001,Skype Minute,AAA,Creator
3,Alex,2001,Skype Minute,B,Author
4,Alex,2001,Contract2,B,Buyer
4,Mary,2001,Contract2,B,Buyer
4,Paul,2001,Contract2,AAA,Seller
5,Mary,2002,Contract3,B,Seller
5,Paul,2002,Contract3,AAA,Buyer
6,Mary,2003,Contract4,B,Buyer
6,Paul,2003,Contract4,B,Buyer
6,Jane,2003,Contract4,AAA,Seller

Another csv example with a header can be found here: got_battles_paoh.csv