"Technique" categorization as voc4cat_0000185(processes) instead of voc4cat_0000181(non-temporal abstract entity)

[HendrikBorgelt]

Most of the "techniques" inside voc4cat are child's of voc4cat_0000185(processes) and thereby a temporal abstract entity.
Which makes them something similar to an event or an action, something that has some time where it originated/started living and some time where it ended.

Most if not all techniques are however discribing something not bound by time. Thus I would switch the classification to voc4cat_0000181(non-temporal abstract entity).

Related terms are:

• modification technique
  • doping as the respective child it is currently describing the process of doping, so maybe this should stay and we should move modification technique to beeing a Collection
• deposition technique
  same here.
  • spray coating
  • doctor blading
  • spin coating
• characterization technique
  all characterization techniques are techniques themself, so no need to turn characterization technique into a collection
  • X-ray fluorescence
  • electron energy loss spectroscopy
  • Raman spectroscopy
  • atomic force microscopy
  • differential scanning calorimetry
  • thermogravimetric analysis
  • nuclear magnetic resonance spectroscopy
  • photoluminescence
  • scanning electron microscopy
  • X-ray photoelectron spectroscopy
  • X-ray diffraction
  • transmission electron microscopy
  • ultraviolet-visible spectroscopy
  • Fourier transform infrared spectroscopy
  • X-ray absorption spectroscopy
    • XAS Transmission
    • XAS Fluorescence
  • electron microscopy
    except maybe for the analysis once.
  • operando analysis
  • in-situ analysis
  • Brunauer-Emmett-Teller surface area analysis
  • Mott-Schottky analysis

This issue mainly occurs due to us not using skos to it's fullest potential, however, except for the fixes propossed above, I don't know how resolve this with a simple fix.
The most correct way to describe them in skos would be to as that all childs of technique are "is_a" child's of respective techniques and are related to process as "part_of" a process.
With collections, we can at least state that a "technique" has_a "(measurement_)process"... To make this statement however correct, we would need to say that the "technique" is a "technique_collection" and may want to define the "technique" term seperately.

[dalito]

I agree with your assessment that the techniques should better be categorized under voc4cat_0000181(non-temporal abstract entity). "BET analysis" and the other "...analysis" may be kept under processes. However, (quite likely) they were also meant to be abstract entities. The contributors may just have used the common term "BET analysis" without thinking about the difference between process and the abstract technique.

A question is if it makes sense to define an additional process of performing a BET measurement in addition to the BET measurement technique. I think for the vocabulary defining both would be acceptable. However, in a data model you would just use one category consistently.

What do you mean by

This issue mainly occurs due to us not using skos to it's fullest potential,

Do you just mean that we do not yet strictly follow our own rules of how to apply SKOS in voc4cat and its top-level categorization?

I would be in favor of changing/fixing this by making the concept definitions clearly reflect if they are "process" or abstract entities and by adding the missing ones with new IDs. By defining both, the "process" and the "abstract entity" we would not break anyones model that relies on the current classification.

Other opinions? @markdoerr @nmoust @RoteKekse @schumannj ?

[dalito]

Interesting: PMDCo treats the techniques only as processes: https://materialdigital.github.io/core-ontology/index-en.html#PMD_0000945 - This should not mean that we should follow. It is just a "FYI" that they do it currently like that.

[RoteKekse]

Hey @HendrikBorgelt @dalito maybe i am too much influenced by the BFO book (by barry smith) and the idea of not using concepts in ontology modelling, but for me all of these techniques are in fact bound in time. Because they describe something which happens. It is not the concept of blade coating, it is the actually doing of blade coating starting with dropping ink on the substrate waiting, using blade to distribute, then maybe some annealing etc.

I think i would prefer changing the term technique into process, i am not really sure what technique actually means if not process.

So deposition process, characterization process (that would be the same as assay in OBI).

[dalito]

@RoteKekse - Thanks for the feedback/comment. - So for you the current state of categorizing the techniques under processes is what you expect/prefer and you would not vote for change (but only for clarification by more precises naming/definition).

[HendrikBorgelt]

@RoteKekse and @dalito , sorry for just posting ChatGPT test in the next section, but I think this explains the problem more suitable than I could:

A measurement technique, such as a GC–MS technique, does not itself unfold in time. It is not something that happens. Rather, it is a specification of how a measurement is to be performed. In BFO terms, such a technique is best understood as an information content entity (ICE): it is generically dependent on some bearer (for example, a document, database entry, or protocol record), and it is about a certain type of measurement process. It provides structured methodological content, but it does not have temporal parts and does not participate in other processes as an occurrent would.

By contrast, a measurement process is an occurrent, more specifically a planned process. It unfolds in time, has a beginning and an end, involves participants, and produces outputs. A concrete example would be a specific GC–MS run performed on a defined sample at a particular time. That run is the measurement process. It has participants such as the instrument, the sample, and possibly an operator, and it results in data as output.

The instrument itself—for example, a GC–MS device—is neither a technique nor a process. It is a material entity (an independent continuant). When it is used in a measurement, it bears an instrument role, which is a specifically dependent continuant. This role is realized in the measurement process. Thus, we must distinguish carefully between:

the material entity (the instrument),
the role it bears (instrument role),
the process in which the role is realized (measurement process), and
the informational specification guiding the process (measurement technique as ICE).

The correction to the current modeling is therefore not to redefine techniques as processes. Instead, measurement techniques should remain classified as information content entities, while measurement processes should be explicitly modeled as planned processes. The relationship between them can then be expressed in terms of conformity or specification: a measurement process conforms to, or is specified by, a given measurement technique. This preserves the BFO distinction between continuants (including material entities, roles, and information content entities) and occurrents (processes), and avoids category conflation.

Going full BFO isn't feasible for Voc4cat, and forcing simple Users to understand an ontology isn't the right approach. However, I want/wanted to reduce confusion with this Issue. I would like to define techniques as techniques and measurements as measurements if possible.

However, and this is coming back to my comment on not using SKOS to the fullest, SKOS has its own ways of dealing with roles, processes, and subclassing, which we could use to partially solve this issue. However, right now, this might lead to more problems than it is worth...

GC–MS Process
    is_a: Measurement Process
    has_a: GC–MS Measurement Technique

would be correct in SKOS.

Now If we want to do this in Voc4cat, our concept GC-MS Process would need to refer to the Collection of GC-MC Measurement Techniques, in which the concept GC-MS-Measurement Technique lives. But we can't really do that with the current voc4cat template can we?

[RoteKekse]

Yes that makes sense, I would agree with chat gpt. And in that sense technique makes sense and is not a concept but something real.

I would then vote for removing technique and replace it with process as a simple solution.

For example blade coating is not a technique as stated. Since the correct execution of blade coating depends on what material is used, what properties are aimed for. For example some blade coating process has quenching as a sub process some does not.

For voc4cat it is more important to model the processes than the techniques. Since there are combinatorial many techniques for just one process.

Same is true for Measurements, eg for XRD, you have 1d, 2d, gracing incident, Bragg-Brentano, you have on powder or thinfilms, different sampel treatment. From my understanding each combination could be a technique then.

[HendrikBorgelt]

Okay, I practically agree, but see another challenge.

I know we are in SKOS, and SKOS does not have instances/individuals in the same sense as OWL, but I intended to use some of the Techniques in a SHACL Schema, in which I would have addressed the Bragg-Brentano version of an XRD Measurement as an Instance of an XRD Measurement Technique for an XRD Measurement device. And since I want to use OBI, I need the ICE, not the process.

I don't know how many people would face similar issues. I would be fine in Missusing a Voc4Cat Process as an OBI:ICE, since we can correct the mistake via a mapping. Changing definitions would, however, take a lot of effort, and currently, basically all definitions and class names would be wrong or easy to misinterpret.

[RoteKekse]

i mean right now they are all in a is arelation with process right, just the parent as the word technique. is it that much effort to rename technique to process and then start a new technique hierachy for this?

[HendrikBorgelt]

i mean right now they are all in a is arelation with process right, just the parent as the word technique. is it that much effort to rename technique to process and then start a new technique hierachy for this?

electron energy loss spectroscopy: An analytical technique used ...
Raman spectroscopy: A spectroscopy technique studying ...
a.s.o. for all of which I created...

and then those that I did not create:
spin coating: A technique used to deposit uniform thin film...
spray coating: A technique used to deposit...
And we can argue about whether a method is a process or a technique, I would rather have it clearly defined without need for discussion, so:
doctor blading: A tape-casting method used to deposit thin films of a catalyst onto the surface of a substrate.

We need more than just a shortcut to fix this, otherwise we will just "refix" this "mistake" we are currently discussing in an other Issue and another PR.

And yes I could implement this in one or two hours. I just want @dalito, @markdoerr, @RoteKekse , your opinions.

Maybe we also need to have failsaves implemented in the definition of process and technique in general, so that people don't get confused about how to classify some aspects.

[RoteKekse]

i am in favor of renaming technique to process and recreating techniques as non-temporal abstract entity for usage as discussed