Re: How do I structure files for ontology development?

continued transcript:

[20:35] <bshambaugh> what was said after, " so I created a file called
x.rdf and a file called x in a directory on my webserver" and before
"it's all implementation-detail minutia" I got kicked off. Sorry poka.
[20:36] <bshambaugh> thanks and happy thanksgiving too
[20:50] <poka> do whatver you want
[20:50] <poka> as long as you conform to content-negotiation per HTTP
RFC (in teh 7200s somewhere)
[20:50] <poka> anything with extensions is a convention but not in the spec
[20:50] <poka> filename extensions that is
[20:50] <poka> trial and error here lead to eliminating symlink usage
[20:52] <poka> with hard links, you dont have to worry about the
symlink's target going away, and it's faster
[20:55] <poka> one thing i could recommend is
[20:55] <poka> install ldnode/nodeld, whatever the solid server from
MIT is called lately
[20:55] <poka> try using their filesystem layout for your daemon, and
if you have issues with it, raise them on the gitter cha
[20:55] --> mischat (~mischat@97e645ae.skybroadband.com) has joined
this channel.
[20:58] <poka> that way, when JAvascript gets dropped by the hipsters
in favor of ES17 with autoclosures and V8 no longer compiles to run
node-solid-server, you can just reuse the same FS layout with your
daemon
[20:58] <poka> although the choice of node does seem like a pretty
good hedge against abandonment right now
[21:00] <-- mischat (~mischat@97e645ae.skybroadband.com) has left this
server (Ping timeout: 260 seconds).
[21:28] <bshambaugh> hardlinking x.rdf with x seems to work to resolve
in the web browser
[21:30] <bshambaugh> I'm not sure how they do an html page at x but
still keep the # URIs at x.rdf relevant
[21:30] <bshambaugh> but I guess that is okay for now
[21:31] <bshambaugh> I'm not sure why I am making an ontology other
than define explicitly what my hash containing URIs mean when using
then in RDF triples
[21:31] <bshambaugh> for linked data purposes for now
[21:31] <bshambaugh> I am not doing any reasoning yet
[21:33] <bshambaugh> thanks for studying ldnode/nodeld solid
whatever...it is on my to do list. It could be my 2008 computer that
is causing the trouble installing. grin, but try I will
-Brent Shambaugh

GitHub: https://github.com/bshambaugh
Website: http://bshambaugh.org/
LinkedIN: https://www.linkedin.com/in/brent-shambaugh-9b91259
Skype: brent.shambaugh
Twitter: https://twitter.com/Brent_Shambaugh


On Thu, Nov 24, 2016 at 8:47 PM, Brent Shambaugh
<brent.shambaugh@gmail.com> wrote:
> Hi all,
>
> I have question about URIs in ontology development. Do terms
> referenced in rdf:about need to be dereferenceable?
>
> For an example, I created a file called isp.rdf at
> http://data.thespaceplan.com/isp.rdf . Each one of my properties and
> classes have the base http://data.thespaceplan.com/isp# . Thus I
> created another file called isp with a full path of
> http://data.thespaceplan.com/isp .
>
> From #swig on IRC:
>
> [Thursday, November 24, 2016] [5:39:02 PM EST] <brent_> I have a
> question about URIs in ontology development
> [Thursday, November 24, 2016] [5:39:11 PM EST] <brent_> I have a file
> called x.rdf
> [Thursday, November 24, 2016] [5:39:47 PM EST] <brent_> and
> rdf:about="x#something" in an xml file
> [Thursday, November 24, 2016] [5:41:06 PM EST] <brent_> so I created a
> file called x.rdf and a file called x in a directory on my webserver
>
> I lost connection, but I saw this:
>
> [20:29] <poka> it's all implementation-detail minutia, but i personaly
> dont use symlinks for doc -> doc.ttl, the symlink resolution actually
> takes a noticeable amount of time on crappy hardware with a bunch of
> files,
> [20:30] <poka> and dumb implementations start returning MIME types of
> application/x-symlink or something for the file, you end up with files
> without extensions which means other SW will try to sniff it
> internally
> [20:30] <poka> plus, it's one POSIX feature you can probably do away
> with if you arent using a FS backend , or are on some miserable OS
> like Windows where a symlink is a "NT Junction" or whatever
> [20:31] <poka> using hardlinks all over the place for index containers though
> [20:31] <poka> which proably makes deletion harder , but i dont
> consider deletion a feature
>
> As an additional question, how would I answer this question if I had
> something like http://data.thespaceplan.com/isp/classname with the
> root
> http://data.thespaceplan.com/isp/ ?
>
> Thanks,
>
> Brent

Received on Friday, 25 November 2016 02:59:19 UTC