search

EBISPOT / ols4

Version 4 of the EMBL-EBI Ontology Lookup Service (OLS)
http://www.ebi.ac.uk/ols4/
Apache License 2.0
39 stars 17 forks source link

"has_dbxref" information is misleading #369

Open sabrinatoro opened 1 year ago

sabrinatoro commented 1 year ago

The "has_dbxref" section on the right side of the page is misleading: it includes ALL the db-refs and there is no obvious way (unless one clicks on the info button) to distinguish which are exact matches and not exact matches. This leads to confusion and misleads users.

Proposal: Show which of the bas_dbxref are exact or not by either showing the "source" information, or by using some flag.

matentzn commented 1 year ago

@sabrinatoro if you hover over the (i) symbol next to the xref, you will see the annotation:

image

I am not sure that making this information more prominent is going to help much, as it is so non-standard (what does "source: Mondo:equivalent" mean?).

dbxrefs are in any case confusing - the fact that people pretend they are "mappings" is really the underlying problem here..

sabrinatoro commented 1 year ago

After discussions, here is an updated proposal: 1) in the "has_dbxref" section, do not include the "has exact matches" 2) next to "has_dbxref", add a warning sign (e.g. triangle with an exclamation point) with a note (e.g that can be seen when one hovers over it) explaining that the precise relationship with these dbxrefs is unknown, and could include narrow, broad, or related matches.

matentzn commented 1 year ago

  1. Has to be an issue on the Mondo tracker!

  2. I think is a good idea. if @henrietteharmse thinks that this would be acceptable, to basically have a warning icon right next to has DB xref stating that:

Database cross-references are not precise semantic mappings. Some of them correspond to exact matches, but many will be broader, or even point to a term that belongs to an entirely different category.
jamesamcl commented 1 year ago

I think also we need to explore how to make the UI for annotations on annotations more first class than just a tooltip (#225), as reification is increasingly a critical part of our plans for expressing mappings. Possibly the annotations would be more of a tree with expandable nodes showing their annotations.

cmungall commented 1 month ago

is this issue still relevant? The axiom annotations on the mondo xrefs were always intended for internal use, there was no expectations OLS builds in specific behavior for this. Having a generic popup for all axiom annotations is sufficient.

We should communicate precise SSSOM mappings using skos