- From: Alberto Reggiori <alberto@asemantics.com>
- Date: Wed, 25 Aug 2004 06:11:35 -0700 (PDT)
- To: public-rdf-dawg@w3.org
[ this use case tries to motivate the requirement to have arbitrary RDF resources representing the BRQL SOURCE part of a query (i.e. not simply a URI) ] MisterX is member of the great FOAF community and he has two different and separated identities or persona(s) [1]; which correspond to his credentials and personalities in two distinguished online communities (let's call them A and B) E.g. he might be member of the RDF community as well as Apache core member and developer - quite separated communities still and he might be important for him to keep his identities separated :) MisterX has three separated FOAF profiles describing its two controversial personalities. The first two FOAF profiles describe MisterX his X persona/character (his email, his chat accounts, his publications and so on) and are kept and maintained separately on different URIs from the the FOAF profile containing his other Y persona information. While MisterX might choose to keep only two separated FOAF profiles one for each of his characters, in reality happens that parts of his FOAF profiles are distributed and spitted over different separated files. For example, he has two online persistent URLs representing each of his X and Y personas, and a third ad-hoc maintained file which contains other information relative about his Y persona (or even he might have a fourth FOAF profile maintained elsewhere containing a mixture of the two personas to facilitate his personal information management). This means that is not generally possible associate a URI (or URL) to his distinguished personas - but instead his characters need to be described, represented and retrieved as separated resources. Assuming that all three FOAF profiles of MisterX are available online, and a given RDF / FOAF harvester program is scuttering them, some source / provenance information about each FOAF URL profile needs also to be kept separately. Such provenance information is explicitly referring only to those URLs containing the three separated profiles and can not be used to pin-point to the two distinguished personalities - but only to their source URL. This generally would allow the crawler to distinguish between them if necessary later on e.g. trackback which URL contains which information. At the same time, each of his two personas are being described in each of the FOAF file explicitly, and his foaf:Person triples are being flagged/scoped into specific characters scopes as well. It is important to note that his FOAF triples (name, email and so on) might also have other contexts/provenance information attached to them. E.g. the URL of the FOAF URL above - where they have been retrieved from. The crawler software can either ad-hoc generate or infer the personas information from running ad hoc queries on the retrieved files - or explicitly (and automatically) associating each RDF triple using some specific RDF parser extension [2]. In other words, the FOAF triples resulting from the crawling process might resides in two (or more) separated scopes/spaces: the URL of the FOAF (source/provenance) and the other, a foaf:Persona typed resource (a bNode in fact) representing his persona(s). Now the crawler has a publish Web interface which can be used by 3rd part services applications. Such service, in addition to provide the raw RDF triple sources also provides all the provenance/context information associated to the triples; such information might be simply embedded into the RDF sources themselves or provided in some ad-hoc generated manifest file and referred from its Web interface. Let's assume that such 3rd party application is a FOAF/people aggregator (e.g. like plink.org) which needs to present and redistribute different people information also globbing RDF descriptions by user persona/character (foaf:Persona) in addition to his foaf:knows or other RDF properties. This means that the 3rd party application has to run ad-hoc RDF queries over the retrieved data to select user specific characters. rdfs:seeAlso: http://rdfweb.org/topic/FoafAndTrust Alberto [1] http://dictionary.reference.com/search?q=persona [2] http://lists.w3.org/Archives/Public/www-rdf-interest/2004Feb/0209.html
Received on Wednesday, 25 August 2004 13:23:14 UTC