Controlled vocabulary for "operationType" contains non-operations

[reckart]

The controlled vocabulary for "operationType" contains mixed up values. E.g.:

• DependencyParsing - the task/process of identifying and marking the grammatical structure of a sentence, establishing relationships between "head" words and words that modify those heads
• DependencyParser - A component that generates a dependency tree from typically token and part-of-speech annotations

IMHO it should only contain the latter (DependencyParser) not the former since we use this vocabulary to determine the type of components.

[reckart]

Or even:

• http://w3id.org/meta-share/omtd-share/MorphologicalTagger - A component that annotates tokens of a text with morphological information (part-of-speech and morphological features)
• http://w3id.org/meta-share/omtd-share/MorphologicalAnnotation - The task/process of adding annotations pertaining to the morphological level of analysis (e.g. gender, number, person etc.)
• http://w3id.org/meta-share/omtd-share/MorphosyntacticTagging - The task/process of adding morphosyntactic tags to words in a text, i.e. part-of-speech and, optionally, morphological features per part-of-speech.

[pennyl67]

As decided in the meetings for ontology, we distinguish between

• operation types for components, for which we use the "agent" style (e.g. parser)
• operation types for applications, for which we use the "activity" style (e.g. parsing) In the ontology, these are clearly distinguished, each having a different broader concept.

In the XSD, although hierarchical relations are encoded, they cannot be shown nor used for validation. The problem is that we use the same element ("function") for both components and applications - in fact, the only difference between them in the XSD is the flag "application" with yes/no values.

In the editor form, when the user selects to describe applications, he is shown only the tree with the values for applications; same thing for components.

So, when writing XML files, one has to be careful in selecting the appropriate values. We can have tips in the documentation but we cannot validate.

[reckart]

We can just as well mark an application which does NER as NamedEntityRecognizer. I see no benefit in assigning NamedEntityRecognition here as the difference between component and application is already encoded in the application flag.

[pennyl67]

The issue has been discussed and decided in one of the meetings for the ontology; there are pros and cons for this decision, but now is definitely not the time to re-open the discussion. Please, for the Maven components, use the values from the component type hierarchy. If it helps, I can send it separately in a json or ttl format. One thing that helps towards your suggestion is the fact that in the ontology (not the controlled vocabulary in the XSD), there's a property "performs operation" which links the two.