search

ESIPFed / sweet

Official repository for Semantic Web for Earth and Environmental Terminology (SWEET) Ontologies
Other
119 stars 33 forks source link

What is the best way for a domain scientist to browse SWEET #226

Open cmungall opened 4 years ago

cmungall commented 4 years ago

Usually when I wish to browse SWEET I load into Protege, but this is often not straightforward, and is not ideal for non-experts

The README has 6 ways of accessing SWEET

Some don't work, e.g for webvowl I get:

image

all of the methods seem pretty low-level and aimed at RDF experts. I'm looking for a simple OLS-type browser.

I'm curious as to why bioportal/ontoportal is not listed (cc @graybeal ). This seems like something I might want to direct a scientist at:

http://bioportal.bioontology.org/ontologies/SWEET

The view is still odd, e.g. I get odd things like nucleon as root:

image

SWEET's design. patterns seem to confound many browsers, e.g. the habit of providing ambiguous labels makes it confusing to search for a hydrological concept like reservoir:

image

There is no guidance as to which should be chosen, it's a bit confusing

also the use of equivalence axioms for synonyms makes the graph confusing. For example, bioportal renders the system concept of reservoir like this:

http://bioportal.bioontology.org/ontologies/SWEET/?p=classes&conceptid=http%3A%2F%2Fsweetontology.net%2FreprSciComponent%2FReservoir#visualization

image

What is not shown is the fact that component and systems-component are part of an equivalence clique. I think this is all a bit hard for an end-user to browse.

Is there a recommended browser for SWEET that takes into account things like non-unique labels and the use of equivalence axioms?

graybeal commented 4 years ago

I imagine many/all of the issues with the links not working are due to the issue you reopened (https://github.com/ESIPFed/sweet/issues/150).

I didn't point users to BioPortal in part because of the view issues you identify. BioPortal was never intended to be an OWL-expert viewer of ontologies, it has always been a relatively naive viewer of hierarchical relations and single-concept annotations (and not all of them either), nor of individuals. Understanding equivalence cliques is way beyond BioPortal's current portfolio. So it may be a particularly poor choice for a domain scientist to view SWEET, in SWEET's current form.

Though I don't think it's fair to blame BioPortal for the existence of ambiguous labels in SWEET. SWEET's decomposed structure has always left it vulnerable to the reality of scientific (or perhaps, non-scientific) naming of terms, that's what the English language does, after all. I don't know what your correct image would show, but I would be surprised if BioPortal has a picture-perfect representation of the ambiguities, even if SWEET used prefLabel and altLabel perfectly. Perhaps a ticket can be created to perform a search of all SWEET for duplicate labels and recommend whether any annotation changes are needed, if that's part of the issue.

Incidentally, your second image is the same as the first, so I suspect it is not the right image.