search

INCATools / ontology-development-kit

Bootstrap an OBO Library ontology
http://incatools.github.io/ontology-development-kit/
BSD 3-Clause "New" or "Revised" License
214 stars 54 forks source link

Make subset the default method for using robot extract #893

Open cmungall opened 1 year ago

cmungall commented 1 year ago

Although SLME guarantees entailment preservation, paradoxically it can lead to ontologies that are more incomplete.

For example, currently CL and UBERON have large confusing incomplete sections in them, see:

I believe the reason GO and others don't have this is because our pipeline bypasses standard extract mechanisms and first makes a super-light version of PR

On top of this, users almost never want the additional terms from SLME. the graph method in SLME is intuitive and easy for people to understand (although of course I would love to move away from the extract-merged owl file being the default and using base modules all the way for anything user facing, but this will take longer)

matentzn commented 1 year ago

I think this issue should be solved by #458, NOT by changing the default module type. Subset is meant for downstream users, SLME is not, but there is no way we will use subset at ontology design time (at least not until we have agreed on burying OWL (and me with it)). We need to ensure logical consistency of GUOBO!