MediaType

[aazocar]

The description of nfdico:MediaType is not clear. The examples are file extensions, in the comment it refers to the file format... but media type could be something like the MIME type as well. Is it something specific or is this an umbrella class for all these mentioned ?

[SashaVsesviatska]

It is a general class, in case you need more specific classes, please introduce them in matwerk ontology.

[saidfathalla]

The problem is that it first makes confusion with object type (media looks like a subtype of object) and second it semantically restricts the type to only media files, not all files, e.g. text files. so, to make it a general class, it is better (from my POV) to rename it to File Type or File format.

[SashaVsesviatska]

Hi, according to the definition, media type and object type do not have subclass-superclass relation, since object type defines the type of an object in a data collection, e.g. photograph, and media type is a specific file format, e.g. tiff.

MediaType is equivalent to the dcterms:MediaType, defined as "A file format or physical medium.", so I do not think we need to create more confusion and define FileFormat and link it to dcterms:MediaType.

I would leave the issue open for more discussion and brainstorming for others!

[saidfathalla]

My point is the use of the word type in two concepts with two different meanings (i.e. photograph and tiff). As you mentioned, media type is a specific file format, so it makes more sense to me to either directly reuse http://purl.org/dc/terms/FileFormat or rename it to media format. At the end, it is your decision, am just trying to improve ;)

[joergwa]

What is the actual purpose of this information?

• To support a developer to render some content (image, video, text, etc.) properly in a website"?
• Support someone who is looking for "videos", "images" or "text-documents" in the dataset?
• Deliver the right content type on HTTP requests?
• ... ?

Again, what is the indended use of "nfdicore:MediaType" ?

[SashaVsesviatska]

To support the search of specific object types in specific formats. MediaType is superclass of FileFormat in dcterms. it basically, additionally to file formats, allows to provide physical description information for an item in a collection, which may be useful in culture domain.

[joergwa]

The definitions in DC are a bit confusing, but reading it from bottom to top, it makes sense for me.

dct:MediatypeOrExtent "A media type or extent." (e.g.: some MIME type or "color lithograph" or "1 minute 30 seconds") ↑subClassOf
dct:MediaType "A file format or physical medium." (e.g.: some MIME type or "color lithograph") ↑subClassOf
dct:FileFormat "A digital resource format." (e.g.: some MIME type)

[hofmannv]

Hm, ..

I think we have a bit of a confusion here. Generally (at least i think) a "type" of something is different from the "format" of something:

• A "type" refers to the content of a thing (i.e. is it spacial data (image, geo-data, etc) or a time-series (audio, voltage recording) or something else (e.g. force curve from tensile-test)).
• A "format" is about how something is encoded (in the digital world). (.jpg; .mat, .whatever).

Thinking on a broad scale (i.e. core ontology), "Media(-files)" should probably be some specific subclass of "files". And what the core ontology would need is a definition of Format and Type for the more general class (i.e. looking for media might actually be more culture specific, whereas in Material Science people search for data types or data formats - which would probably also be some kind of "files" (or some other named or already existing useful superclass)).

Not sure if the subclass hierarchy of dct:MediatypeOrExtend makes really sense, because there is some correlation between formats and types, but no strict causal relationship (i.e. i can save pictures in many different formats .jpg, .svg, .raw etc..)

In my eyes the core ontology would benefit from using e.g.

• cdt:Format to define a format (i.e. also for any type of files and
• dct:type to define a type (note also that the comment points to the use of format. As properties on a general level. Things like "Media-files" should then inherit these properties through the hierarchy.

That would then also allow

To support the search of specific object types in specific formats.

I would be curious what you think...

[joergwa]

@hofmannv

A "type" refers to the content of a thing (i.e. is it spacial data (image, geo-data, etc) or a time-series (audio, voltage recording) or something else (e.g. force curve from tensile-test)).
A "format" is about how something is encoded (in the digital world). (.jpg; .mat, .whatever).

dct:FileFormat might fit the second case ("format") E.g. a MIME-Type (also dct:MediaType might fit, because it is a superclass of dct:FileFormat.) (then, also nfdi:MediaType fits because it is equiv to dct:MediaType)

For the first case ("type"), I understand that you are referring to some specific (materials) science related types of content. This does not fit the definition of dct:MediaType ("A file format or physical medium."). Here we need something else. Is there some controlled vocabulary?

@aazocar In the ontology nfdi:MediaType (equiv to dct:MediaType) refers to "A digital resource format or physical medium.": I adapted the description and added some more examples. I also introduced a MIME-Type class below nfdi:MediaType.

[saidfathalla]

we discussed this issue again:

• the cdt:Format is the formats defined by the list of Internet Media Types which is not exactly the file extension we want.
• So, we find that it is better to define a new DP called fileExtension which we can use to specify the file extensions of an image dataset, for example.

[joergwa]

What do you suggest as domain for the DP fileExtension?

[saidfathalla]

Oh sorry, it's my mistake. it should be any Resource.

[joergwa]

OK. added.

[joergwa]

@saidfathalla One more question about the fileExtension property In the Culture Graph I found (most) media types are used together with Wikidata. See here

How do you plan to use fileExtension as DP ?

Would you mind to suggest a description for the DP and how it differs from the mediaType property in usage. So that users better understand when to use which?

[saidfathalla]

We are planning to use fileExtension for specifically the file extension like txt, or any non-standard extension like xyz which can be equivalent to wikidata:file extension but media type is different (A file format or physical medium), i.e a MIME type or "color lithograph", as you said above which could be text/csv.

A def. could be: An identifier for a file format (e.g. txt for a text file) used as a suffix to the file name, don't use dot at start.

[joergwa]

Thank You!