2007/5/12, Evgeny Egorochkin <<a href="mailto:phreedom.stdin@gmail.com">phreedom.stdin@gmail.com</a>>:<div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> On Saturday 12 May 2007 18:02:08 Mikkel Kamstrup Erlandsen wrote: > > While > > > this could be done, the machine readable ontology does have quite a few > > > benefits. Fx: > > > > > >  * You could update the ontology without updating any applications or > > > > search > > > > > engine code > > > > > >  * 3rd parties could extend the ontology by installing their own ones > > > > Yes, you could update or extend the ontology, but the new fields won't be > > automatically populated until the engine is told how to get those from > > the original data. > > Does this make sense ? > > Yes you are entirely right. There are still many reasons to allow for > extensibility though. > > It might be that some search engines will extract more rich data than the > xesam spec. With easily introspectable fields applications can pick this up > on the fly. > > It has also been discussed several times how to use common metadata > extractors. If applications could install a special-purpose extractor then > we also need a way to define new fields... The idea is that applications will use Xesam API to introspect fields, so to them it doesn't matter where field definitions come from, which format etc. And the question is maybe we should use API to define these fields instead of text-files?</blockquote><div> Well, the current ssearch API contains no way to introspect the availble fields (or set them). This could be part of a metadata API, which is yet to be discussed. Under any circumstance I think that it is valuable to be able to parse and update the ontology without a running service... </div> <blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">like xesam::field f("music:composer"); f.addParent("content:author"); </blockquote><div> Would this be a dbus service or just a library? I don't think it is a good idea to hide the actual implementation behind a library. In my eyes the Right Way (TM) is to standardize the way to create xesam ontologies and then provide a helper lib to do it (if it is necessary at all). </div> <blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> So each meta-data extractor defines their specific fields via API + xesam core defines a set of pre-defined fields to enforce the standard.</blockquote><div> Yes, I think that xesam should provide a ready-to-ship ontology. How to install new ontologies is another matter, how to install new extractors is yet another matter that has only been discussed lightly without much consensus... </div> Cheers, Mikkel </div>