- From: Young,Jeff (OR) <jyoung@oclc.org>
- Date: Wed, 22 Sep 2010 10:26:51 -0400
- To: "Monica Duke" <m.duke@ukoln.ac.uk>, "public-xg-lld" <public-xg-lld@w3.org>
Monica, Here are some thoughts on protocol aspects of degrees of openness. I wouldn't expect all Linked Data to be open data. If the Linked Data resources aren't open, though, I hope the URIs secure themselves using standard HTTP mechanisms. Here are some use cases: - If you don't want to reveal anything at all to unauthenticated/unauthorized users, then have the URIs return 401 (Unauthorized), 403 (Forbidden), 404 (Not Found) or 407 (Proxy Authentication Required) status. - If you are willing to reveal "free beer" to unauthenticated/unauthorized users, then create dual HTML and RDF representations: one each for authenticated/authorized users and one each for unauthenticated/unauthorized users. For example: http://example.org/person/1 [300 (See Other) redirect to...] http://example.org/person/1/ [200 (OK) content-negotiate to...] http://example.org/person/1/open.html (unauthenticated/unauthorized HTML representation) http://example.org/person/1/open.rdf (unauthenticated/unauthorized RDF representation) http://example.org/person/1/closed.html (authenticated/authorized HTML representation) http://example.org/person/1/closed.rdf (authenticated/authorized RDF representation) (Using "303 URIs forwarding to One Generic Document" <http://www.w3.org/TR/cooluris/#r303gendocument> will help preserve SEO performance by keeping users focused on the generic resource URI. Don't forget to set the Content-Location header on the generic resource.) - Crawlers can be deterred by using 503 (Service Unavailable) and an escalating Retry-After header. - If you make the RDF available in a resolvable named graph, secure the graph using HTTP status codes as described above and assert licenses and restrictions against the graph's URI. - Include license and restrictions assertions in the RDF for each individual. (Maybe just on the individuals identified with 303 URIs?) - Leave the resources open, but put the server behind a firewall (e.g. make it easier for employees to do their jobs without giving away the farm.) I don't have any suggestion for which license/restriction ontologies would be best for library resources. Jeff > -----Original Message----- > From: public-xg-lld-request@w3.org [mailto:public-xg-lld- > request@w3.org] On Behalf Of Monica Duke > Sent: Wednesday, September 22, 2010 6:35 AM > To: public-xg-lld > Subject: [Fwd: Re: on expressing access constraints in a data > repository of mixed openness] > > The following query has come up on another mailing list (for discussion > of data management issues in research) and although the question refers > to repositories it made me wonder if this might not also apply to some > of the library linked data? Or is there an assumption that all > (linked) > data is going to be made available on an open basis? > > I've had a (quick) look at VOID [1] but I could not spot anything that > fits the request. The closest is the recommendation for announcing the > license of a dataset. For automatic analysis it is recommended to use > "canonical identifiers for well-known licenses" examples given include > creative commons licenses and GNU. Perhaps something similar is needed > for sharing availability? > > Is such a need likely to arise for library data, and if so are there > existing methods to communicate the availability or is that a gap that > the XG could highlight? > > Regards, > Monica > > [1] http://vocab.deri.ie/void/guide > -----Original Message----- > From: Research Data Management discussion list > [mailto:RESEARCH-DATAMAN@JISCMAIL.AC.UK] On Behalf Of Chris Rusbridge > Sent: Monday, September 20, 2010 5:17 AM > To: RESEARCH-DATAMAN@JISCMAIL.AC.UK <mailto:RESEARCH- > DATAMAN@JISCMAIL.AC.UK> > Subject: On expressing access constraints in a data repository of mixed > openness > > [Apologies for cross-posting.] > > I'm looking for some more help. I'm hoping that at the very least the > discipline of writing down my concern will help me understand it > better, > and at best you guys might have a solution. > > Let's imagine an institutional data repository (which I guess could be > a > set of different repositories). By definition, the IDR will have data > that have different degrees of openness. I can distinguish at least > these conditions: > > a) fully open > b) closed until some condition is met (then to be open) > c) closed unless some condition is met > d) closed indefinitely. > > I'm not really sure an IDR would actually want to accept data with > condition (d), but there may be good reasons that escape me at the > moment. But however much one would like all data to be open, there are > substantial swags of data that must be temporarily or partially closed. > > Independently of conditions (b) to (d), it is possible that some or all > of the metadata might be open, that is to say the data might be > discoverable even if not open (presumably if you found and wanted to > use > the data, then some sort of negotiation would have to take place). > > My question is: how could constraints like these sensibly be expressed, > in either a human-readable or (better) machine-readable way? > > -- > Chris Rusbridge > Mobile: +44 791 7423828 > Email: c.rusbridge@gmail.com <mailto:c.rusbridge@gmail.com> > > >
Received on Wednesday, 22 September 2010 14:28:17 UTC