Posted by: Anonymous
on August 26, 2008 08:19 PM
I've been following Nepomuk for a while now and it does indeed seem promising. The KDE integration is great, and expanding all of the time (just follow Planet KDE for about a week and a few new uses will likely be blogged about). Of the whole Nepomuk project itself, I'm afraid I have some doubts about the scope. There are many papers published in the Deliverables section of their site, but the integration systems apart from KDE (Mozilla for example) seem dormant or at least stalled, and there are more ideas being incorporated (such as Gnowsys and SELF) which seem admittedly related but don't really seem to fit.
KDE's Nepomuk integration is very pragmatic; infrastructure has been built and is expanding, use cases are appearing and being implemented. More developers would be great, but it is obviously going ahead and picking up pace. As for other Nepomuk systems, they don't seem particularly grounded in real-world applications. Integration is the key, since I would not spend time in a Nepomuk frontend adding metadata and constructing ontologies if the only use of them would be inside the Nepomuk frontend. I would vertainly do it if it were in my music player (Amarok), regular file manager (Dolphin), image viewer (Gwenview), etc. but also if it were in my browser (the Mozilla integration, as I say, seems dormant), in my messenger, in the websites I visit, etc.
RDF, SPARQL and OWL are standards and thus future interoperability is pretty straightforward. Get Nepomuk plugins/integration with WordPress, LiveJournal, Firefox, the GIMP, Inkscape, Abiword, OpenOffice, identi.ca and as many Free Software projects as possible which deal with authoring and publishing. Make it seamless to add metadata, not as an obscure, hidden menu item. Then push for Nepomuk integration with services like Google, Flikr, DeviantArt, CreativeCommons.org, feedburner, YouTube and as many other content-pushing services as possible.
If metadata (automatic and manual) can be preserved/expanded upon from drawing a picture in Inkscape, uploading it to Flikr, being downloaded by someone else and made into a slideshow video posted onto YouTube with music from Newgrounds, with the final video containing all metadata from its constituent parts then it really would be a semantic web. The thing is, this level of permeation is emergent it cannot be planned. It relies on action, useful code and willingness to help integration efforts, not necessarily debating the merits of peer2peer topologies or other such details. That will come in time as the initiatives pick up momentum.