here are some quick thoughts, after a quick experiment converting a foaf 
file to json.   It seems to me that one huge value for json over rdf-xml 
is human readability, and we ought to try to preserve that.  So:

*(minor) I'd suggest pretty printing the json: indenting the properties 
of each item, and starting a newline for each property
*(minor) For similar reasons, I would drop quotations where they are not 
necessary (eg around property names with no special characters)
* (major) the converter uses the fully qualified property names from the 
rdf as the property labels in the json output.  this makes for 
unweildy/unreadable property names.  I think it would make sense to 
apply the heuristic of taking the final fragment (after the last slash) 
of each property as the "human readable name" (which would correspond to 
the exhibit "label") of the property, and reserve the fully qualified 
property name as a uri (or exhibit "id") of the property.  Although I'm 
not certain exhibit currently supports labels and ids for 
properties---davidh?  In any case, dropping the uri and just keeping the 
trailing name would probably be worthwhile in many cases, even though it 
involves a bit of information loss (could create false sharing of names). 

Maybe all these thoughts should be for a different output format: JSONH 
(human)?

* (selfish) the effort to have the json have 100% fidelity with the RDF 
means that literal values, instead of being just literal values, are 
actually being created as sub-structures with value:the-value and 
type:literal.  For exhibit we type property ranges instead of individual 
items, producing a more compact and readable though less-expressive 
representation (it prevents properties from having value of multiple 
types).  It would be nice, in cases where the input fits this restricted 
format, to use the more readable representation (and would make exhibit 
happy).

Keith Alexander wrote:
> Hi folks,
>
> I'm working on a semantic format conversion service[1], much along the  
> same lines as babel. It uses ARC[2] under the hood to parse RDF, eRDF,  
> RDFa and microformats, and offers Exhibit JSON and JSONP as output  
> formats. Thanks to David Huynh for helping me to get the exhibit html  
> output working :)
>
> If anyone has any suggestions on how to make the conversion better  
> somehow, or improve the generic exhibit html output, they'd be much  
> appreciated.
>
> Cheers
>
> Keith
>
> [1] http://convert.test.talis.com
> [2] http://arc.semsol.org
>
>   
_______________________________________________
General mailing list
[email protected]
http://simile.mit.edu/mailman/listinfo/general

Reply via email to