Search API

[Nahuel]

Hello,

We are coding a software for social network analyse, we try to mesure
the impact of science journals in those social networks. One of the
networks we want to use, is the Zotero one, that is used save
bibliographic references.

The code of this project will be fully open sourced, probably under
GPL.

For this, we were looking for a search API, to search for all the urls
we host (we are doing this for the revues.org platform).
At the moment, there is no API like this in Zotero's server
API(afaik).
We think this is a great lack of the server API, because it can be use
by a lot of platform to search related articles, by subject, authors,
and anything else.
Of course this kind of service must be anonymized for privacy
protection.

Are you able to receive contributions? We can find time to develop
this part, using a SolR/Lucene backend.

Thanks for answering,

Bests,

--
Nahuel ANGELINETTI

[Dan Stillman]

Can you provide some more detail on your vision for this API, ideally
with (rough) example queries and the sorts of data that would be
returned? I'm not sure we'd benefit from implementation help on this,
but we've always planned global search and have some work towards that
underway internally, so depending on what exactly you're looking for we
may be able to offer something fairly soon.

Obviously, as noted in the forum thread, there are also privacy issues
that would need to be resolved.

- Dan

[Jean-André Santoni]

Hello Dan,

For now, we are in the prototyping phase of our project so we only
need a very simple service that returns a score of an URL:

GET http://yourapi/score?url=<percent encoded:http://mydocumenturl>
28

Where 28 would be the number of users that added the document, or any
other way to judge of the interest of the users for a document.

As you can see, our needs are on a very precise point. I don't know
Zotero well yet, so I can't have a clear vision of what can be served.
Like Nahuel said, we would use solr to serve as much metadata as
possible, with some computed fields to serve statistics about the
ressource, (like the "score" that we need).
Side services can be used to provide global stats, or time driven data
to draw curves, it all depends on how data is stored.

Let me ask my colleagues, they may have a precise idea of what the
ideal API would be.

- Jean-André

[Nahuel ANGELINETTI]

Hi,

Le 12/07/2011 10:01, Jean-Andr� Santoni a �crit :

> Hello Dan,
>
> For now, we are in the prototyping phase of our project so we only
> need a very simple service that returns a score of an URL:
>
> GET http://yourapi/score?url=<percent encoded:http://mydocumenturl>
> 28
>
> Where 28 would be the number of users that added the document, or any
> other way to judge of the interest of the users for a document.
>
> As you can see, our needs are on a very precise point. I don't know
> Zotero well yet, so I can't have a clear vision of what can be served.
> Like Nahuel said, we would use solr to serve as much metadata as
> possible, with some computed fields to serve statistics about the
> ressource, (like the "score" that we need).
> Side services can be used to provide global stats, or time driven data
> to draw curves, it all depends on how data is stored.

Datas like "subjects" can be verry usefull, but this kind of datas
should be in metadatas provided.
The choice of solr is because we are experienced, and its known to be
really "fast". Other great feature in solr is its schema easily extensible.

Bests,

--
Nahuel ANGELINETTI