Re: [JSON] one data point: Anzo on the Web

On 3/18/2011 7:10 PM, David Wood wrote:
> Hi Lee,
>
>> On the wire, we serialize RDF in JSON, of course
>
> Why "of course"?  Why not use Turtle or n-triples or something else?

For a JavaScript client? Because I've got about a hundred JSON parsers 
to choose from (we use whatever comes with dojo). Why choose Turtle or 
n-triples?

Lee

>
> Regards,
> Dave
>
>
>
>
> On Mar 18, 2011, at 10:38, Lee Feigenbaum wrote:
>
>> (I keep repeating that I'm more interested in the other areas of our group's work than the JSON bits, but I keep coming back to discuss the JSON bits. Oh well.)
>>
>> My organization offers a software product called Anzo on the Web, which is, among other things, a faceted browser, BI tool, and data collection tool that runs inside a Web browser. It works against our Anzo semantic middleware, and as such is all driven by SPARQL queries and RDF data.
>>
>> Architecturally, Anzo on the Web is made of 3 layers:
>>
>>   * Anzo.js -- this is a low-level library that communicates with the Anzo server and exposes various APIs for Web apps to be built upon.
>>   * Anzo Web Toolkit(AnzoWT) -- this sits on top of the Anzo.js library and acts as a Web application development framework. It adds easy ways to access RDF data and visualize it (declarative path-based data binding) as well as other app framework bits (toolbars, menus, layouts, etc etc)
>>   * Anzo on the Web (AotW) -- this is the end application that users interact with. It's built using the Anzo Web Toolkit.
>>
>> At the highest level, end users interact with the AotW GUI. At the next level down, developers interact with AnzoWT, manipulating RDF via paths and, if they want to, dipping down to the Anzo.js level to work with SPARQL queries and graph APIs.
>>
>> On the wire, we serialize RDF in JSON, of course. We use a flat, quad-based approach very similar to both the SPARQL Result Set binding and http://www.w3.org/2009/12/rdf-ws/papers/ws02. Anzo.js turns this simple serialization into graphs, etc. for upstream manipulation.
>>
>> So, we would be supportive of work to standardize a simple serialization for triples and quads in JSON, as it would open the possibility for our Web framework to interoperate with other RDF middleware / RDF databases.
>>
>> But, as I've said a lot, it's not by any stretch of the imagination our highest priority.
>>
>> Anyways, just a data point,
>> Lee
>>
>> PS While we don't use it that often, we do have functionality within the Anzo.js library that provides easy object-style access to the RDF data. We don't do it via simple JSON properties, but instead more Java-style with getters and setters. See http://roubenmeschian.com/rubo/?p=19 for some examples. Again, this is  something that we generate via our libraries by performing processing upon the simple, flat, RDF-in-JSON serialization.
>>
>
>

Received on Friday, 18 March 2011 23:27:17 UTC