Issue/What metadata to record
< Issue
On retrieving an about record, the client get certain information. Here a collection of the types of information that various database registries have recorded. We need to decide what we want to record. I've partitioned these as to whether they are specific to the record or to the databank.
At the record level
- A record type. There is often some effort say what the record is about - the subject. However there is little consistency in how that is done. LSRN typically records one type. NCBI doesn't record any. MIRIAM can have several. Uniprot uses a system of categories. Vocabularies can be controlled or uncontrolled.
- Pointers to the encodings (us)
- Format of the id in some other format, e.g. as NCBI would cite the id
- Historical, but now dead, links
At the databank level
- long name, for display purposes (most). For names and other language dependent strings, expect internationalization.
- A short name (all)
- Alternate names (miriam)
- A descriptive summary of the databank
- Home page for the databank (uniprot,lsrn,miriam)
- A pattern that matches the id (better - a resource of active ids) (all)
- A url template for each encoding (all)
- Historical, but now dead, versions of pattern and url_template
- Rights/License information regarding use of the record (freebase)
- Link to Help or Documentation (miriam)
- Database schema, if relevant (freebase)
- Provider search page (freebase)
- Citation of publication about the databank (uniprot)
- Responsible institution (miriam)
- Location of instituion (miriam)
- Location to download the records in bulk (encoding specific)
- Mark of obsolescence (miriam) Implication: We don't ever retire databank records.
- Link type (implicit or explicit) (uniprot)
- A pattern to match to see whether the page was not found: e.g.ATCC : "You’ve reached an old or invalid URL for ATCC.org"
- A list of other databanks that links to this databank
- A list of other databanks that are linked from this databank
- A list of known encodings for this databank
In all cases
- Change notes
- Provenance information - who supplied each piece of information (OWL 2 lets you do this at the entity or axiom level)
- DOAP (Description of a project) record for Shared Names
