Vocabulary: http://www.ivoa.net/rdf/datalink-core Author: Baptiste Cecconi Date: 2021-06-09 Term: #detached-header Action: Addition Label: Detached Header Description: Machine-readable metadata for #this, which in general will be necessary for its scientific use. Examples include FITS headers distributed without their data blocks or PDS label files. Relationships: rdfs:subPropertyOf(documentation) Used-in: http://voparis-tap-maser.obspm.fr/voyager_pra/epn/dl/dlmeta?ID=VG1-J-PRA-3-RDR-LOWBAND-6SEC-V1.0%3APRA_I.TAB Term: #documentation Action: Modification Description: Structured or unstructured metadata helping to understand, interpret, or work with #this. Such information can range from processing logs to weather reports to technical documents on instruments to related publications. Rationale: In some formats and archives, the metadata required to decode the content of #this is in a separate file. In the case of NASA/PDS data products, for example, the PDS label file contains the decoding metadata. In some archives, the FITS header may be stored separately as a plain text file, next to a data file consisting of a binary stream of bytes. Clients would use the content type (MIME type) (e.g.: application/x-pds4-label+xml, or text/x-pds3-label) to enable the processing. As to the change of the definition of #documentation, is essentially dropping the previous requirement of it being "human readable". On closer inspection it turned out that humans will always need tools to interpret bitstreams, and hence "human-readability" is not a useful and reproducable criterion to define concepts. Perhaps at some point we will have to tell apart structured and non-structured metadata so computers can ignore non-structured documents. As long as detached headers are the only sort of structured metadata computers look for in datalink documents, there does not seem to be sufficient reason for introducing this distinction. Discussion: In the thread at http://mail.ivoa.net/pipermail/semantics/2021-June/002806.html, the proposal was received favourably. There was a proposal to perhaps add a wider term *#metadata, but given there already is #documentation which probably covers a large part of such a concept already, no immediate need was brought forward. On the other hand, the discussion brought up that #documentation explicitly asked for "human-readable text form". This certainly is not the case for many important sorts of detached headers, and even for FITS headers one can debate this. In consequence, the VEP was amended to also fix #documentation's definition (cf. http://mail.ivoa.net/pipermail/semantics/2021-September/002846.html). The VEP was accepted in the present from at the TCG session of 2021-10-27.