To design the metadata structure, a set of goals that this metadata should serve is to be established. Based on these goals and available information about platforms, instruments and the like, a prototype for the metadata structure will be established. This document is explicitly meant to be updated, comments are always welcome.
The achievement of these goals should be easily decidable by a yes / no answer.
- establish ownership and responsibility for measurements, instruments and platforms, e.g.,
- association with a PI
- institutional ownership, if applicable
- contact for usage questions
- provide means of acknowledging use of data.
- ability to localize measurement (time, place, instrument, quantity), e.g.,
- through unique names / ids to platforms and instruments
- list of measured quantities
- platform that hosted measurement
- support the cataloguing of larger datsets, e.g, for use by other tools, such as intake.
- index to storage locations of measurement data (may also be generated from metadata included in data files)
These goals should be strived for, but the outcome of this might not be easily decided on by a yes / no answer.
- machine readable
- community sourced
- should evolve incrementally
- make things more findable
- ability to trace version history of measurements
- ability to trace version history of metadata
- ability to identify commonalities with other measurements, e.g., from the same or similar instruments
- provide a means to share common analysis conventions, e.g., segmentation, thresholding
- adopt existing standards to the extent possible, e.g., GCMD Keywords. See also here and here
Things which are explicitly not a goal of the metadata structure
- focus on filenames as ordering scheme
- enforce a single tree structure
- imposed preconceived ways of working or sharing of data