How to encode preferred labels for a particular source

[matentzn]

Scenario: You have a group that says: I prefer my disease to be named X, rather than using the primary label (rdfs:label). There are basically six ways to do this:

1. 🚀 Use skos:preferredLabel with dc:source
2. 👍 Use oio:exactSynonym with oio:hasSynonymType
3. 👀 Use rdfs:label with oio:inSubset
4. 👎 Use rdfs:label with dc:source
5. 🎉 Sub-annotation property of IAO:'alternative term', i.e. 'nico preferred term' subPropertyOf iao:'alternative term'
6. 😄 SKOS XL (see @alanruttenberg comment below)

All of these have problems, most of them have churn being the main issue (the amount of work needed to standardise this). However, I would suggest that we at least try for some kind of standardisation.

• https://github.com/Southern-Cross-Plant-Science/cdno/issues/64
• I am against 1 because preferredLabel intuitively seems to be singular (there should not be multiple "preferred" labels). Also, we don't make that much use of preferred label property in OBO, so it seems a lot of churn to updated our tooling (NLP etc) to consider these.
• 3 is the conceptually cleanest, but suffers from the greatest churn - a lot of our tooling assumes that there is a single label.
• 4 uses only standard properties, but it seems a bit weird, I have never seen it. Also, same problem as 3.
• As exactSynonym is ubiquitous in OBO now, and people have been using these synonym types in various pipelines already, I feel that that option 2 is the way to go. The main argument for 4 is the arguments against the rest. Also I like the synonym types and would like to do much more with them in the next years.

[graybeal]

Do I understand correctly that exactSynonym is creating not just a new label, but a new concept/IRI? If so, that creates a different kind of churn which is potentially more impactful, as there is no single well-defined way for someone (or some computer) with that IRI (who may not have the memo about its annotations) to find the mapped object. So the meaning could get entirely lost.

My apologies if I should know how oio:exactSynonym worked, I tried looking it up on the webiverse and gave up after some minutes. Where can I find the OIO that you're using?

[bpeters42]

I think there is a #5 approach (used in OBI) where we create sub-annotation-properties of 'alternative term', such as 'IEDB alternative term' (http://purl.obolibrary.org/obo/OBI_9991118). The benefit of this is that within the IEDB website, we know that we want to always use these labels. Within our system, they are precise and short. But we do not think at all that they are good labels when viewed outside of the IEDB context.

On Mon, Jun 21, 2021 at 9:59 AM John Graybeal @.***> wrote:

Do I understand correctly that exactSynonym is creating not just a new label, but a new concept/IRI? If so, that creates a different kind of churn which is potentially more impactful, as there is no single well-defined way for someone (or some computer) with that IRI (who may not have the memo about its annotations) to find the mapped object. So the meaning could get entirely lost.

My apologies if I should know how oio:exactSynonym worked, I tried looking it up on the webiverse and gave up after some minutes. Where can I find the OIO that you're using?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/information-artifact-ontology/ontology-metadata/issues/69#issuecomment-865195226, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADJX2IVN7XKW67FU26WDJULTT5VVBANCNFSM47BCXNKA .

-- Bjoern Peters Professor La Jolla Institute for Immunology 9420 Athena Circle La Jolla, CA 92037, USA Tel: 858/752-6914 Fax: 858/752-6987 http://www.liai.org/pages/faculty-peters

[allysonlister]

FWIW, Although it's not an OBO ontology, our subject ontology as used in FAIRsharing (https://github.com/FAIRsharing/subject-ontology) also uses the OBI method of using sub-annotation-properties of 'alternative term', and it has worked very well for us.

[alanruttenberg]

And a 6th: SKOS-XL https://www.w3.org/TR/skos-reference/skos-xl.html, in which labels are reified. I think you get most of the benefit if you use #5 and pun the relation and an individual. You need to do that if you explicitly want to link a representation of IEDB with the label property. With that you could say 'IEDB alternative term' is_preferred_label for 'IEDB web site',

Alan

On Mon, Jun 21, 2021 at 11:14 AM bpeters42 @.***> wrote:

I think there is a #5 approach (used in OBI) where we create sub-annotation-properties of 'alternative term', such as 'IEDB alternative term' (http://purl.obolibrary.org/obo/OBI_9991118). The benefit of this is that within the IEDB website, we know that we want to always use these labels. Within our system, they are precise and short. But we do not think at all that they are good labels when viewed outside of the IEDB context.

On Mon, Jun 21, 2021 at 9:59 AM John Graybeal @.***> wrote:

Do I understand correctly that exactSynonym is creating not just a new label, but a new concept/IRI? If so, that creates a different kind of churn which is potentially more impactful, as there is no single well-defined way for someone (or some computer) with that IRI (who may not have the memo about its annotations) to find the mapped object. So the meaning could get entirely lost.

My apologies if I should know how oio:exactSynonym worked, I tried looking it up on the webiverse and gave up after some minutes. Where can I find the OIO that you're using?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub < https://github.com/information-artifact-ontology/ontology-metadata/issues/69#issuecomment-865195226 , or unsubscribe < https://github.com/notifications/unsubscribe-auth/ADJX2IVN7XKW67FU26WDJULTT5VVBANCNFSM47BCXNKA

.

-- Bjoern Peters Professor La Jolla Institute for Immunology 9420 Athena Circle La Jolla, CA 92037, USA Tel: 858/752-6914 Fax: 858/752-6987 http://www.liai.org/pages/faculty-peters

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/information-artifact-ontology/ontology-metadata/issues/69#issuecomment-865206225, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAB3CDSIYXC7OFKFZSXVDPDTT5XPBANCNFSM47BCXNKA .

[matentzn]

@alanruttenberg who is using labels like that? I have never heard if this before.. Actually I was not even aware of SKOS XL - its interesting but very far away from anything I have seen so far in OBO. I added it to the options up and above, but I guess we would need some concrete examples.

@bpeters42 thank you, I have added your suggestion to the voting above! It also looks like a good idea. You save one level of indirection (annotations on synonym assertions) which is good. I guess I was sort of hoping the days of IAO:'alternative term' are numbered in favour of oio:exactSynonym, but probably not :D

@graybeal Exact synonym does not create a new instance, at least not a named one. Basically you say:

x oio:exactSynonym "A" [oio:inSubset some iedb:IEDB_SUBSET]

where both oio:exactSynonym and oio:inSubset are annotation properties.

[alanruttenberg]

On Tue, Jun 22, 2021 at 2:38 AM Nico Matentzoglu @.***> wrote:

@alanruttenberg https://github.com/alanruttenberg who is using labels like that? I have never heard if this before.. Actually I was not even aware of SKOS XL - its interesting but very far away from anything I have seen so far in OBO. I added it to the options up and above, but I guess we would need some concrete examples.

A google of skos xl filetype:owl pulls up some uses. Vivo, it appears. Agree that we haven't used in OBO. I didn't intend my mention of it to be an endorsement, just as another potential technical solution. A concrete example would be to represent community-specific labels with different instances skos-xl:Label.

@bpeters42 https://github.com/bpeters42 thank you, I have added your suggestion to the voting above! It also looks like a good idea. You save one level of indirection (annotations on synonym assertions) which is good. I guess I was sort of hoping the days of IAO:'alternative term' are numbered in favour of oio:exactSynonym, but probably not :D

@graybeal https://github.com/graybeal Exact synonym does not create a new instance, at least not a named one. Basically you say:

x oio:exactSynonym "A" [oio:inSubset some "IEDB"]

Is the "A" a string or rdf:type? Having trouble parsing this.

where both oio:exactSynonym and oio:inSubset are annotation properties.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/information-artifact-ontology/ontology-metadata/issues/69#issuecomment-865739262, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAB3CDXMQOVYZUQYLJWHUHDTUBDXHANCNFSM47BCXNKA .