Submitted by Mandriva has sent out a press release to highlight NEPOMUK, the semantic framework (think metadata) being developed by Mandriva and several partners and will be integrated in the upcoming KDE 4. It includes a link to a video demonstrating NEPOMUK integration into Dolphin, the KDE 4 file manager. In addition, “aKademy 2007 has kicked off! The first weekend hosted our user conference, which brought many talks about various topics, ranging from very technical to more practically oriented, which were spread over two tracks. The tracks were interweaved with keynote talks. Read on for the report of the aKademy 2007 keynotes.”
This looks very promising. I assume, I can search the tags and ratings with some kind of desktop search, e. g. Beagle. Maybe this brings us a step closer to a non-pedestrian way of managing our data. Nice.
I really wish people would tone down the crazy names. They have NOTHING to do with what the program or protocol does and they aren’t even that great sounding.
Nepomuk is not a crazy name — or at least, it’s not invented by KDE. It’s the name of the FP6 funded EC research project into the semantic desktop. (I’m not sure why they choose Nepomuk — this particular saint is much more suited for on-line forums, since he’s a patron saint against calmunies. Or perhaps they just took the town name.)
“””
Nepomuk is not a crazy name — or at least, it’s not invented by KDE.
“””
Well, it is a bit questionable. Not the worst ever, to be sure. ( http://tinyurl.com/2qryg2 )
It’s too bad that they did not decide to spell it backwards. Then it would have started with a ‘K’ and had the string ‘open’ in it.
“It’s too bad that they did not decide to spell it backwards. Then it would have started with a ‘K’ and had the string ‘open’ in it. :-)”
It would also sound like a bad porno movie.
I don’t want to be a spelling czar but it’s aKademy.
Nice to see metadata (or file attributes) arriving on linux desktop, at last!
Nothing specific to kde4, but look at the video, what’s with this 2 seconds delay when you click on a directory? I know this is beta stuff, but current linux file managers are way too slow imho, i like it fast!
Actually, after watching the video, the delay is less than a second. No doubt it will be faster when a) it’s not alpha code, and b) it’s not recording a full-screen videocast at the same time.
Linux file managers are not particularly slower than what you’ll find on other OSes.
Well, you’re right about the videocast and alpha code, and i wasn’t trying to make a point with this video, but that’s something i see every day, for years, under gnome, kde, winxp, and even beos.
From my experiences under xp, explorer stalls sometimes (start menu too), and it’s very irritating, but the rest of the time, traversing directories is almost instantaneous. Traversing directories under Gnome is slower than windows on the same hardware. Etc etc, there _are_ differences, and it should be analyzed.
Sincerly i’d like to see a serious benchmark on this, that would measure those stalls and delays under each platform. I’m sure Haiku would be well noted, even if it’s pre alpha code
True, there are differences, but the OS isn’t the only factor. If you’ve got previews turned on in KDE it slightly slows down navigation in Konqueror.
You’re right, this should be analyzed…you should do it.
(Not being crass or anything…if you don’t do it, maybe nobody will!)
“You’re right, this should be analyzed…you should do it. ”
well, i may
Odd, I find that GNOME and KDE file managers are faster than Explorer on Windows in most cases.
Admittedly, seem to recall that Windows seemed faster on a clean install, but it seems to slow down for me after about 2 months or so to the point where it’s quite a bit slower. It’s still in the order of 1/10 second, but still noticeable.
OS X Finder seems to have the fastest general speed, but overall the other FTFF-type of bugs (icons appearing ontop of each other, stupid arrangement and crap) makes it slower for me to use it. Also, Finder isn’t great at huge directories.
Nice demonstration, it doesn’t show much , but like many will comment, it looks promising. The tagging feature in itself is,for me, one of the most important ways of organising your data.
It’s like the taxonomy feature in Drupal, allowing for ad hoc organisation of information. With multi user access to files this will make collaborative work a lot easier in the absence of the more usual collaborative tools,also provides a nice base framework for existing tools. It’s so exciting because the tools which can evolve from here would be revolutionary.
It’s an exciting development with many big industry players behind the project ( i,e IBM , SAP ,HP et al). Now if the NEPOMUNK services are provided through a daemon, using DBUS for the messaging layer then it should hopefully be a DE Agnostic Solution. I’m really excited about this.
Hmm can’t hardly wait till pclinuxos integrates all this into their usual tight KDE implementation.
kudos to all who make the development possible.
Edited 2007-07-05 12:40
Do they really spect we tag our files?
What’s the innovation in this? Vista is already doing it.
Where are those automatic tagging algorithms and semantic linking and so on that I heard about a few years ago?
“””
Where are those automatic tagging algorithms and semantic linking and so on that I heard about a few years ago?
“””
In the next major version of Windows, of course.
Most of these semantic frameworks (currently working together to provide a common interface via the XESAM project at freedesktop.org) also feature file content indexing, which allows you to search for, aggregate or modify files based not only on their file attributes (name, size, etc.), but also on their content and metadata.
Also, Vista fails to realize most of the potential that tagging brings. You can’t tag (or search within, for that matter) arbitrary files in Vista, only certain supported formats. The open source semantic frameworks provide a modular interface for their indexers and a central store for tags and file-external metadata. Because of this openness and modular design, it’s trivial to search for (or within) any filetype you’d like.
Nice but, I though the goal of NEPOMUK was to avoid all the manual tagging, it was sold like something super innovative, but all I see is something that have already been done before.
Edited 2007-07-05 17:07
I can understand why people would want to rate their music, but not necessarily their files. I can only understand rating in comparison to other files – for example if I stored several documents in a folder relative to a project I was working on, I would want to score each file based on relevance, content quality, detail etc against all the other files in that folder. With this, I can only give each file a rating of x number of stars. What I’d like to do is rate each file, on a per tag basis.
That is the only way I can think of rating generic files (i.e. not media) being useful to me – having a rating per tag would MASSIVELY improve how I deal with my data. Just using an arbitrary scale of 1-5 is not that much use to me; I want to know how useful a file is going to be and, more importantly, I want my desktop search to understand that relationship too.
Edited 2007-07-05 15:17
I agree. It seems to me that many devs are not thinking that much about what all the meta-data is going to be used for.
Really, the only useful implementation of meta-data handling I’ve seen so far is the ID3-tagging interface in Amarok which is excellent and works on nested data so that you can quickly re-write and clean up your heterogeneous music collection. Hopefully this ability will be there in Nepomuk/Dolphin as well.
Moreover, what happens with file-system meta-data when files are shared between systems? One of the few reasons I can come up with concerning rich tagging usage is to tell other people how I perceive some data. How does Nepomuk achieve this kind of persistence?
Edited 2007-07-05 17:41
Its funny, when I first read about the upcoming DB filesystems across the major platforms, I got really excited. FINALLY we will be able to do away with the hierarchical folder structure that has plagued pretty much all interfaces since the dawn of time. Finally we can see some new objects being created outside of the folder/file/shortcut trinity. Finally file names will become only a minor identifier of a file, and we can start using more efficient descriptors.
Well, we have Windows Search, Spotlight, and Beagle, and what has that given us? Basically locate on steroids.
Nepomuk is one of the first signs we have of someone actually upping the bar. Granted, this is a very early stage, but IMHO it shows more potential then any of the other big players.
Isn’t this (nepomuk) one of Sebastian Trüg’s project’s?
I think he is at least the head developer. He has done a fantastic job with K3b. I hope nepomuk turns out just as useful.
http://liquidat.wordpress.com/2007/05/31/semantic-desktop-and-kde-4…
Yep, Sebastian is indeed the Mandriva lead guy on Nepomuk.
just as a P.S. – Seb also won this year’s Best Application aKademy Award for k3b
The name Semantik was already taken
http://www.kde-apps.org/content/show.php?content=55242
the data accumulated via nepomuk is searchable via strigi. the two work together to bring a graph backed by an rdf store and full text searching.
tagging is the most basic and naive usage of this technology, though that already is useful. they are working towards a full social/semantic desktop which will bring some pretty neat stuff. at the same time, everyone knows that we don’t really know everything people will figure out to do with it. we have some good ideas of what we can do, but the most interesting things are probably waiting to be dreamed into being by some future hacker once they have kde4 in hand.