rename experimental modality -> experimental approach?

[tgbugs]

@UlrikeS91 @lzehl @satra for many people modality => data modality, whereas experimental approach is more nebulous (which we want in this case since approach conflates data modality, technique, paradigm, etc.). This would avoid the overlap in terminology, but we will need to update our communities on the preferred terminology (in this case the label of the superclass). Thoughts?

xref openMetadataInitiative/openMINDS_controlledTerms#36

[satra]

@tgbugs - i think we had this discussion a bit when you presented techniques and modalities. and we talked a little bit about whether modality could be inferred. what are thoughts on not having modality at all? are there example cases where modality/approach could not be inferred from other metadata?

[tgbugs]

In principle I think all of them can be inferred from the more granular techniques, but we would still want the classes as controlled terms. The original use case was to be able to provide a more general/accessible layer for categorizing datasets in search, so even if they were no longer required as part of deposited metadata, we would still want to retain them in the ontology. Also sometimes it is more straight forward for data depositors to provide the asserted modality/approach, which we can then use as a sanity check. Practically I also tend to view it as something of a backstop when granular information is missing/not ready yet.

[lzehl]

@satra 🙂 I had the same question, or actually asked where the modality/approach stops and the super-classes of techniques begin. But as @tgbugs said having them as controlled terms to categorize data seems indeed to be often required for searches. I personally like to change it and rather use modality in context with data (@tgbugs: you mean with "data modality" here, e.g., time-series, correct?).

[UlrikeS91]

Having modalities at all: I think it is good to keep them as controlled terms as well. I agree with all of you. They could be inferred from other metadata but we should keep them for e.g. searches and sanity checks (when more granular techniques are available).

Renaming modalities to approach: I'm all for it. I've gotten used to modalities (and other curators too), but I like approach better. It's more suitable and should also be easier for naive users to understand. I remember that I didn't get what was excepted for "modality" when it was first introduced as a field for our datasets, only after I saw the options I could choose from.

[satra]

pinging @yarikoptic so he knows this is happening.

i'm fine with changing the name, but we should have a bit of a conversation around this, as we have been using a set of "modalities" to create suffices for our files on DANDI and there is a mapping based on the class hierarchy within NWB schema.

[tgbugs]

Definitely would want to understand the impact for that, is there somewhere in the dandi code that I can take a look to assess the impact? The only changes should be to the naming, the structure will otherwise remain the same, and I'll switch to use the opaque InterLex identifier so that people won't be confused by ilxtr:ExperimentalModality.

[tgbugs]

I've made the switch, holler if anything goes wrong.