[Accessforall] Thoughts on Metadata

Andy Heath andyheath at axelrod.plus.com
Mon Feb 6 21:25:22 UTC 2012


This may need an agenda item to discuss.

I have an action item due completion end of Feb to list/summarise 
Metadata efforts - this may be wrong but is how I understand it right 
now anyway.  However, I haven't yet taken any action to move it forwards 
and its worth explaining why. Let me first state that I am using the 
term Metadata to mean "stuff about data" not stuff about people - so 
preferences are not Metadata in this usage (not everyone uses this 
definition).

I've been thinking about what needs doing here and come to the 
conclusion that driving anything from the Metadata end in this work is 
not the right way to go.  My feeling on this is that we should *only* be 
building preferences architecture and content (and *some* uses of 
preferences) into the basic structure of the standard.  I agree 
Gottfried's proposed approach to do this on attribute-value pairs each 
pair representing a preference is a good way to do it (sorry if I 
labelled it wrong) and that any relevant Metadata approach should be 
"mapped" *to* that or built on top of it and not the other way around. 
Further there are a *huge* range of Metadata schemes out there, all tied 
up with the politics of organisations that use them or that make money 
from them and its an error to allow the preference work to get entangled 
in arguments around those or their politics.

Even further, my thinking is

1. that there will be many schemes/structures built on top of the 
preferences and they won't necessarily work together or even factorise 
(mathematically I mean partition the space).

2. That all work related to Metadata (and probably all work related 
things like device properties) should probably come from or have a 
*very* strong relationship with the marketplace - which means vendors of 
content and vendors of platforms.  Perhaps there should be *some* 
influence by the very largest of organisations using Metadata - but this 
is very very messy - we would imho really want to get those players on 
board not be doing it *for* them.  Well they already *are* on-board 
aren't they (I'm thinking Cloud4All) ?  I don't know what vendors are in 
Cloud4all.  I don't see any point at all in designing ideal Metadata 
approaches that will then be ignored by the marketplace chasing its own 
business cases (this *is* what we have been doing for years). We are in 
an environment where business cases are moving very fast, particularly 
around publishing and I see it as essential that we are working *with* 
business cases that are in active use or that vendors are developing 
right now. This might mean ignoring existing Metadata schemes completely.

So what would I do ?

a. for now, just ignore Content Metadata Schemes and develop the 
preferences (ensuring only that techniques to implement each preference 
are available *somewhere*)

b. take those preferences to vendors (e.g. Cloud4All) and develop *with* 
those vendors the needed Metadata schemes in concert with their 
implementation

c. bring those schemes (and possibly parts of implementations) back into 
the standards as examples of how the preferences could be used in 
particular scenarios (but not the *only* way the preferences can be used).

I hope this makes some kind of sense and I'm happy to elaborate/discuss 
on the call or the list.

Cheers

andy
-- 
__________________
Andy Heath
http://axelafa.com



More information about the Accessforall mailing list