Annotation details

[James Teters]

Just curious if there is any documentation on how annotations will be
implemented?

Any ideas on size limitations or restrictions for this meta data?

[Raffi Krikorian]

we will be releasing data in due time!

--
To unsubscribe, reply using "remove me" as the subject.

--
Raffi Krikorian
Twitter Platform Team
http://twitter.com/raffi

[Nigel Legg]

I look forward to reading about this, sounds... intriguing.

[Jud]

On Apr 14, 5:05 pm, James Teters <jtet...@gmail.com> wrote:
> Any ideas on size limitations or restrictions for this meta data?

good question; I have the same one.

simple math based on average tweet status byte size (of status
structure coming through the streaming or REST interface) tells us
that it wouldn't take much being jammed into the annotation's field to
double that size. what status size increase is Twitter's
infrastructure ready/willing to tolerate?

it seems to me that a few things are NOT candidates for the
annotations field(s):
- void * (for you old schoolers on the list)
- media who's original native format is binary (e.g. photos/videos)

annotations will need limitations like:
- overall size
- if key/value pairs become the model... they'll need individual size
limitations (for name and value)
- max number of pairs
- etc.

the whole thing feels driven by the answer to the original "size"
question.

another question would be whether or not the tweet originator can
remove annotations that others put on their tweet? I'd assume that I'd
have control over my original tweet in that manner (e.g. "notes"
functionality on Flickr)

[Raffi Krikorian]

we're not sure on the sizes we are going to initially launch with, but i suspect we will launch with something small and ramp it up.  i think the limit will also be the "total sum of stuff", as opposed to "number of keys", "length of keys", etc.

--
To unsubscribe, reply using "remove me" as the subject.

[Dewald Pretorius]

Raffi,

Is the planning for everyone's annotations to be available to everyone
else, or will there be private namespaces accessible only to the
source application?

[Rajiv Verma™]

From my Twitter Client, which is a desktop one, (rtweetz!), when I tweet & I have any & symbol, it simply skips tweeting.....y??

--
Thanks & Regards
Rajiv Verma
Bangalore
E-Mail: raji...@gmail.com
Ph: +91-92430-12766
Go Green, Use minimum natural resources!

[Raffi Krikorian]

completely public, except for protected tweets in which case they would be private.

--

To unsubscribe, reply using "remove me" as the subject.

[zn...@comcast.net]

----- "Jud" <jval...@gmail.com> wrote:

> --
> To unsubscribe, reply using "remove me" as the subject.

In addition to size constraints, I'd like to *strongly* suggest that wherever possible, annotations use *existing* open standards! Please, let's not "reinvent the semantic web", even if we can. ;-)

[Raffi Krikorian]

please feel free to point us to standards that you would like us to consider.  we are really attempting to make this insanely simple by literally just having a triple of items to store (namespace, key, value) -- so, we are just really talking about representation, i assume.

[Michael Bleigh]

Will annotations be indexed and searchable? Will I be able to search
for all tweets with a certain annotation namespace, or namespace:key?
I think this would be key to truly creating agreeable standards for
metadata that can be utilized by many clients.

On Apr 15, 9:05 am, Raffi Krikorian <ra...@twitter.com> wrote:
> please feel free to point us to standards that you would like us to
> consider.  we are really attempting to make this insanely simple by
> literally just having a triple of items to store (namespace, key, value) --
> so, we are just really talking about representation, i assume.
>
>
>
>
>
> On Thu, Apr 15, 2010 at 12:09 AM, <zn...@comcast.net> wrote:
>

[M. Edward (Ed) Borasky]

I'm thinking of something like the RFC process for Internet protocols.
By the way, on a related note, once the "Twitter link shortener" I've
been hearing rumors about is in place, can we have all the links in
tweets sent from the API shortened with it? Profile images, user
object URLs, etc. ;-)

Part of this stems from my concern over something I thought I heard
yesterday about Twitter building its own "place" database. There are
dozens of place databases - why does Twitter need another one?

On Apr 15, 6:05 am, Raffi Krikorian <ra...@twitter.com> wrote:
> please feel free to point us to standards that you would like us to
> consider.  we are really attempting to make this insanely simple by
> literally just having a triple of items to store (namespace, key, value) --
> so, we are just really talking about representation, i assume.
>
>
>
> On Thu, Apr 15, 2010 at 12:09 AM, <zn...@comcast.net> wrote:
>

[Raffi Krikorian]

Will annotations be indexed and searchable? Will I be able to search
for all tweets with a certain annotation namespace, or namespace:key?
I think this would be key to truly creating agreeable standards for
metadata that can be utilized by many clients.

the plan is yes - we will be working with the search team to make that happen.

 

[Raffi Krikorian]

I'm thinking of something like the RFC process for Internet protocols.

really - i think that's just too formal.  just mail the list, or hit me/marcel up over email.

 

Part of this stems from my concern over something I thought I heard

yesterday about Twitter building its own "place" database. There are
dozens of place databases - why does Twitter need another one?

honestly, of all the place databases out there, none of them fit our needs.  none of them have the combination of unrestrictive licensing + data and IDs for countries going down to neighborhoods (arbitrarily sized things) + have the ability for creation, updating, etc.  we are building something that will be available through the API that the entire ecosystem can use (and, not just for tweeting), so its a fairly unique set of constraints.

 

[André Luís]

+1!! ;)

On Apr 15, 2010 7:09 a.m., <zn...@comcast.net> wrote:

----- "Jud" <jval...@gmail.com> wrote: > On Apr 14, 5:05 pm, James Teters <jtet...@gmail.com> wro...

[André Luís]

Thanks for that info. Will try to gather a few and send them later.

So you're ruling out concepts w/ multiple properties? Like a vcard?
This seems similar to what axschema.org have for openid. Namespaces have to be uris, obviously.

Cheers,
André Luís

On Apr 15, 2010 1:09 p.m., "Raffi Krikorian" <ra...@twitter.com> wrote:

please feel free to point us to standards that you would like us to consider.  we are really attempting to make this insanely simple by literally just having a triple of items to store (namespace, key, value) -- so, we are just really talking about representation, i assume.

On Thu, Apr 15, 2010 at 12:09 AM, <zn...@comcast.net> wrote: > > > ----- "Jud" <jval...@gmail.com...

[Mathias Herberts]

> honestly, of all the place databases out there, none of them fit our needs.
>  none of them have the combination of unrestrictive licensing + data and IDs
> for countries going down to neighborhoods (arbitrarily sized things) + have
> the ability for creation, updating, etc.  we are building something that
> will be available through the API that the entire ecosystem can use (and,
> not just for tweeting), so its a fairly unique set of constraints.

Hmm doesn't OSM contain sufficient data to actually be turned into a
place database?

I'm thinking administrative boundaries et al.

Mathias.

[André Luís]

Why shorten links that won't count for 140 limit and are not viewed by user? It will only add un-needed requests and waste values on the twiter shortener.

André Luís

On Apr 15, 2010 2:18 p.m., "M. Edward (Ed) Borasky" <zzn...@gmail.com> wrote:

I'm thinking of something like the RFC process for Internet protocols.
By the way, on a related note, once the "Twitter link shortener" I've
been hearing rumors about is in place, can we have all the links in
tweets sent from the API shortened with it? Profile images, user
object URLs, etc. ;-)

Part of this stems from my concern over something I thought I heard
yesterday about Twitter building its own "place" database. There are
dozens of place databases - why does Twitter need another one?

On Apr 15, 6:05 am, Raffi Krikorian <ra...@twitter.com> wrote:

> please feel free to point us to standards that you would like us to > consider.  we are really att...

> On Thu, Apr 15, 2010 at 12:09 AM, <zn...@comcast.net> wrote: >

> > ----- "Jud" <jvale...@gmail.com> wrote: > > > > On Apr 14, 5:05 pm, James Teters <jtet...@gmail....

[Raffi Krikorian]

a way to think about this is analogous to geo.  people used to put geo information in the 140 characters -- but now, we allow you to put it out of band in a machine-readable way.  we want to extend that functionality to all types of meta data (links to URLs, etc.).

[M. Edward (Ed) Borasky]

I guess I need to look at the "protocol buffers" spec again. And some
of the "binary JSON" formats. While we're dreaming, how about sending
Streaming data *compressed*? ;-)

On Apr 15, 7:49 am, Raffi Krikorian <ra...@twitter.com> wrote:
> a way to think about this is analogous to geo.  people used to put geo
> information in the 140 characters -- but now, we allow you to put it out of
> band in a machine-readable way.  we want to extend that functionality to all
> types of meta data (links to URLs, etc.).
>

> 2010/4/15 André Luís <andreluis...@gmail.com>

[M. Edward (Ed) Borasky]

My impression was that the Open Street Map project was attempting to
solve this. At least that's what I picked up in the aftermath of the
Haiti earthquake. If you haven't already, check out http://maps2.humaninet.org/
and http://www.humaninet.org/maps2/maps2-geo-usability-2010-1-12.pdf

We've got a couple of really sharp open source mapping geeks in PDX -
try @GeoPDX and @elsewisemedia for starters.

[Cameron Kaiser]

> I guess I need to look at the "protocol buffers" spec again. And some
> of the "binary JSON" formats. While we're dreaming, how about sending
> Streaming data *compressed*? ;-)

How about keeping a new way of talking to Twitter human readable during its
initial implementation? Premature optimization.

--
------------------------------------ personal: http://www.cameronkaiser.com/ --
Cameron Kaiser * Floodgap Systems * www.floodgap.com * cka...@floodgap.com
-- In Computer Science, we stand on each other's feet. -- Brian Reid ----------

[Raffi Krikorian]

probably not - we're just going to stick with JSON and XML for a bit now.

[Raffi Krikorian]

tell them to hit me up.

--

To unsubscribe, reply using "remove me" as the subject.

[Joseph Cheek]

Wow, another cool idea introduced at Chirp (was it?). Honestly, I'd
just love a free (beer) geocoding service that didn't force me to use
google maps.

Twitter has helped me make $10 so far (woohoo!); perhaps as I make more
with it I'll actually be able to afford going to Chirp! 8-)))

Joseph

--
Subscription settings: http://groups.google.com/group/twitter-development-talk/subscribe?hl=en

[kidehen]

All,

Re. annotations has anyone seen this example re. compact 3-part
(tuple) statements:

http://semantictwitter.appspot.com/

Kingsley