bioimage-io / bioimage.io

Website for the BioImage Model zoo -- a model zoo for bioimage analysis.
https://bioimage.io
MIT License
53 stars 18 forks source link

Ontology for the tags #309

Open oeway opened 1 year ago

oeway commented 1 year ago

It would be nice to suggest tags based on existing ontology with our own complementary tags.

As a reference implementation, we have already the CI script to generate a owl file from a spreadsheet here.

This allows us to use the EBI ontology service api which works by sending HTTP request to the EBI server and return suggested tags from the ontology.

@fjug @akreshuk @k-dominik @arrmunoz @esgomezm

beatrizserrano commented 1 year ago

I've started the mapping of the current tags with the EDAM Bioimaging Ontology during the BioHackathon 🙂 I've found ontological terms for some resource_categories, need to check with @matuskalas (who is here too).

This is what I have so far:

bioimage.io resource_categories EDAM Bioimaging Ontology Comment
modality Topic -> Imaging
dims Data -> Image (or Data -> Image characteristics -> Image data properties?)
content Data -> Region of interest Not sure about this one though, it's complicated since this can be a cell, a tissue (covered by UBERON), or the whole organism (covered by NCBI taxonomy)
framework Could be taken from ITO?
software
network
task Operation -> Image processing
akreshuk commented 1 year ago

Maybe I don't understand how ontologies work, but would it be possible to leave out the top levels of this hierarchy? Everything we have will be under "Topic->Imaging" and "Operation->Image processing". To be clear, what I would like to avoid is people having to select the only relevant entry from some enormous drop-down list, so I don't object to having the mapping as such, I just don't want to expose users to more choices than they strictly have to make.

beatrizserrano commented 1 year ago

That's what I think. We can take only the relevant "subtrees" to give users the choice between the categories defined there.