This application definition, is it for raw or processed data?

[prjemian]

A question that recurs from users contemplating which NeXus application definition is most appropriate for their particular case:

How do I know if a given application definition is for raw or processed data?

Can we make this more obvious for each application definition? Is there an alternative way to provide this information without adding an additional attribute to the definition element in the schema?

[phyy-nx]

Discussion from Code Camp:

• Raw data usually means totally uncalibrated which isn't always useful or even available. What about original?
• NXprocess indicates the data is processed, for example in NXsqom, the processed results are added to the file as NXprocess groups. This means a "Raw" data application definition can serve as a union of raw and processed data.
• Some examples are clean: NXtomo and NXtomoproc, which follows a convention of adding 'proc'
• Other examples don't follow that convention: NXmx is for the original data, NXreflections is for processed data. But NXreflections communicates a lot to crystallographers, so the different name is highly useful
• Fuzziness creeps in when it is unclear if processing is applied to raw data, or if the data is modified or supplemented in a workflow
• Suggestion from @Gigarrum: sort the Application definition page by obviously raw/original vs. obviously processed. Workout fuzziness on case by case basis. Maybe add headers to the top of the two lists.

Proposal from the floor: add a link to Processed data description near the top of the Application Definition page, then supplement the Processed data description with more documentation.

[woutdenolf]

I volunteered to take the issue but imo there is no point in making a distinction between raw and processed data. Application definitions provide a data processing interface to any data, regardless of the origin/history.