Author: Tutte Institute for Mathematics and Computing
A repo for using the TIMC vectorizers and thisnotthat libraries to construct a joint vertex/hyperedge embedding and exploration.
In this demo we will make use of the recipe hypergraph dataset. It consists of recipes (hyperedges) that are sets of vertices (ingredients). Each recipe is assigned to a country (edge label). Here is a short data description.
- number of nodes: 6,714
- number of hyperedges: 39,774
- number of edge label categories: 20
- rank of hypergraph (maximum hyperedge size): 65
The end-goal is to provide a way to explore the data, cluster and classify.
The recipe data can be found here:
- mamba env create -f hypergraphs-simple.yml
- conda activate hypergraphs-simple
- python -m ipykernel install --user --name hypergraph_simple --display-name hypergraph_simple