mnets
Hypergraph Interchange Format invitation
Hi everyone, I'm Martín from NetSci2025, we had some discussions with @arashbm and @dataspider :). I was hoping to follow up with you on a standardization opportunity using HIF, The Hypergraph Interchange Format that I presented at the conference.
The HIF standard defines a schema for well-known network types, currently abstract simplicial complexes, hypergraphs and directed hypergraphs. I think extending HIF to support pymnet's multilayer types would make the standard much more valuable. Also, extending pymnet's netio to support HIF would also mean seamless interoperation with between this and many other higher-order libraries.
Let me know what you think. Cheers!
Thank you for the follow-up, @colltoaction ! It seems to me that extending the standard to handle multilayer networks would be fairly straightforward, but it might make sense to streamline the representation of "multilayer types" in HIF across all participating multilayer network libraries (feel free to object @bolozna @arashbm ). What process did you have in mind for developing the extension (we could add parsing support for HIF on our end afterwards, as the multilayer semantics is key to our library)?
On a different note, if you want to make HIF into a community-driven project, I'd suggest moving its repo ownership to a GitHub organization to signal its independence from any individual account (we set up pymnet this way, too)...
https://github.com/pszufe/HIF-standard/issues/25
I think in Mikko's Multilayer networks paper, in section 2.6 "Hypergraphs", we have a good starting point. If we define the straightforward mapping between the two network types, more restricted ones such as multiplex and temporal can be further refined. Or viceversa, we can start with a small use case and leave multilayer for later.
What process did you have in mind for developing the extension?
It is two-fold: all libraries are invited to adopt the HIF standard as it is today. For example, you can add parsing support to take a hypergraph file and construct it as a multilayer network if it meets the expected constraints, or else fail or display a warning.
On the other hand, extending the standard with types such as "multilayer" and "temporal" would be great. I think more people will be interested, so I can reach out to start the conversation. In this case I hope two libraries would put the proposed changes to the test.
I'd suggest moving its repo ownership to a GitHub organization
I agree! Thanks for your quick response and feedback.
Hi all, I wanted to give you a heads-up on the availability of the Hypergraph Interchange Format paper on the arXiv. Cheers!
http://arxiv.org/abs/2507.11520