Re: Common Tag - semantic tagging convention

Hi Francsois & others,

I really like the turn this debate has taken! Practical considerations
about what is useful and what not!

Going with Common Tag standard through many iterations, I think I can
explain some choices taken.

First, we were aware of other existing possibilities for semantic
tagging. The idea was to approach this from the other side - from what
companies are already doing with semantic tags and what they wanted to
do next. So we were asking ourselves: "can this be made interoperable?".
We tried to create a specification that would be minimal, yet useful for
practical aims of already existing applications and services.  

The parties involved didn't see the immediate need to exactly specify
the the one who placed the tag, but we did feel the need for the ability
to specify how tag was created (inferred by machine, fully by human,
etc...) So we included that. 

The date of tagging came into the game mostly as an example of how the
standard can be extended if the practical need arises. I don't think any
of the people involved actually publish tagging date along the tags
right now. And I agree with Peter Mika, that this can be quite important
piece of information
[I can imagine one reason why taggingDate can be technically very
important - because vocabulary entity (in DBpedia for example, I think
less so in Freebase) can change meaning through time. While this was
definitely not the reason to add tagging date as part of the
specification, it can be justified that way]

As Jamie said, this is the basic skeleton that was needed, now we'll see
how it is used and adjust accordingly through time.


bye
andraz

On Fri, 2009-06-12 at 19:13 +0200, François Dongier wrote: 
> 2009/6/12 Peter Mika <pmika@yahoo-inc.com>y t
>         Maybe others can comment as well, but I do think it's
>         [taggingDate] an important piece of information, e.g. to
>         determine recently popular tags. 
> 
> In my very humble opinion, **who** tagged a resource with ctag T at
> time t could also be a very useful information to store. I'm expecting
> that in the near future we will have rich user profiles, based on sets
> of semantic tags (semantic tagclouds, if you prefer). Communities of
> interest, not just individual people, could also be defined in terms
> of such semantic cIouds. I think lots of interesting computations
> could be done over that kind of information: personalised reading
> recommendations obviously, but also relativisation of the popularity
> of a tagset (and I agree the timestamp is useful for that) to a
> particular community of users.
> 
> 
> For this sort of thing, don't we need a "taggedBy" property?
-- 
Andraz Tori, CTO
Zemanta Ltd, New York, London, Ljubljana
www.zemanta.com
mail: andraz@zemanta.com
tel: +386 41 515 767
twitter: andraz, skype: minmax_test

Received on Sunday, 14 June 2009 05:21:56 UTC