[SPEC] Backwards-compatible metadata in Gemini

[Lars Noodén]

On 2/24/21 9:16 PM, PJ vM wrote:
[snip]
> ...  unless you know that the set of articles in
> which you're searching is actually using that convention,
> and in a consistent way.
[snip]

It is a matter of scope and it is why the use of document metadata is
generally found only in-house for local search and indexing
functionality:  Only you have control over your own documents and the
presence and quality of their metadata.  What goes into the metadata is
a different topic than the specification for encoding it.  So it is
possible for groups with common sets to collaborate.  Or not.

In-house, there is no substitute for the ability to browse or search
specific fields when retrieving from a pool of in-house documents.  A
simple specification for how to annotate key-value pairs is all that is
needed.  Nothing more.

A major benefit from standardized markup for key-values is that any
server-side software for indexing, browsing, and searching is easier to
share.  Think htdig but for a Gemini capsule.  Through the discussion
here I am starting to realize the importance of having a possibility for
capsules to choose whether to add metadata to their own documents.
Otherwise, there is a very low upper limit on the number of documents
they in practice are able to manage and make discoverable.

/Lars