The Semantic Desktop Wants You (KDEDot)

What is left is to produce software that can do something more useful with
the information it creates.

it works...

I alias tracker-search to sch

$ sch vim
Defaulting to 'files' service
Results: 16
/home/drag/Repo/stuff/vim/plugin/taglist.vim
/home/drag/Repo/stuff/vim/plugin/minibufexpl.vim
/home/drag/Repo/stuff/vim/plugin/bufkill.vim
/home/drag/Repo/stuff/vim/plugin/supertab.vim
etc etc.

Without it the best comparable would be something like:

find /home/drag -type f -print0 |xargs -0 grep vim

Of course tracker takes a fraction of a second to return results versus a
10-20 minutes with find/xargs/grep

And it is useful for more then just keeping track of the first 2500 words
of contents of files...

When applications take advantage of the api to index their data, do the
connections and tagging, and use the connections in helpful ways, it will be
useful.

The chromium browser uses the google data store to respond when you type
into the address bar. Very useful, but all the technology is hidden.

If Nepomuk is used in the same way, it will become indispensable.

Derek

> 2. Beware the paperclip. "I see you're trying to watch a video! Can I
help you with that?"

The idea is more to integrate it into applications, so you can easily get
at files you're working at using a specific project, or things you've
edited, or read lately. Another result will be an implementation of
contacts on your desktop. These will be defined in standardized way, using
for example the PIMO ontology, a benefit here is that applications can
more easily use the same concepts, for different goals and have their data
shared. Things like Recent Documents become more reliable, and more
powerful, tagging and rating of files becomes universal across your
desktop and applications. I think this kind of tangible benefits is often
missed when talking about The Grand Vision behind Nepomuk.

Modern desktop OSes leak information like f-ing crazy. There is no way around
it. This software just uses that to your advantage. Not using it is not going
to hinder any sort of experienced investigator in any way.

If you want to protect yourself in a sane and actually useful manner then
encrypt your harddrive using LUKS and shut it down when your not using it.
Then use encryption to protect information that is especially sacred so that
if a attacker gains access to your computer during runtime then they will not
likely be able to access it.

Quite the opposite, it sounds very good for Linux. BEsides, KDE runs on other systems besides
Linux...

"we do not need some nosey piece of software crusing the system looking thru all your files
making a note of just what is where"

Who is this "we" you talk about? What makes you think that you are in a position to talk about
other people besides yourself?

I use both Linux and OS X. And on OS X I have Spotlight which indexes my HD and makes my
contents easy to find. And I at least find that very very useful. Nepomuk seems to take that idea
one step further. This is about making it easier for you to use your computer. If you think that
this is some kind of tool for authorities to snoop on you, then I think you are way off-base.

"easy meat for the FED's ect not a good idea at all we must have the choice of enabling or
disabling it COMPLETELY ie completely removed from the system and that choice needs to be
loud and up front with a FULL explanation of exactly what it does where it looks the sort of
information it stores and Where it stores that information so that it can be completely removed
from the system without trace and by that i mean full forensic traces ie absolutley no chance at
all of anyone ever being able to find it ."

Have you thought about wrapping your computer in tin-foil? That should also help. Seriously:
there's really no need for paranoia here. And hey, if you are so paranoid, why are you online? Do
you have any idea how much information can be gathered from you based on your net-use?

And have you heard of these things called "periods"? They are thing in your keyboard that looks
like this: . Have you considered using them in your text?

I find it surprising that some people have powerful computers (and just about all computers are
powerful these days), yet they insist on leaving all that power unused and use software which does
not take advantage of that power in any shape or form....

If you are stuck on Pentium II or something then go ahead and run something like LXDE. Their
system-requirements seem to mirror Windows 98. So if your computer is 10 years old, it might be
smart to not run KDE. But have you ever consider getting a faster computer? Seriously?

Sure, dreaming, we're not there yet. But the technologies are quickly
maturing. Akonadi is ready for real use now, and with KDE 4.4 apps will be
using it - 4.5 might see a wholesale migration. Nepomuk - exactly the same
story.

Not unusual, but a whole bunch of data that needs to be kept around.
Previously we would have file folders with all these documents, filling
boxes in the back room. With that system, it is easier to go to the site
and read the serial number off the equipment as opposed to digging through
a box of files to find some obscure piece of paper.

It is possible to build an application to keep track of all this stuff. A
database with binary globs, all sorted, indexed and the like. But what I
need to do is different from what you need to do. And I'm not in an
industry with huge support from software vendors. There is choice, but all
of them assume a way of managing the business that may not be my way, and
may need a bunch of people in the office to keep the whole thing current.

So what if the stuff was sorted and indexed? Not simply a per word index,
but on ontologies or structures of specific data?

What if my application using the indexing and ontology of nepomuk (or
similar) to not move around, file or whatever these disperse sources of
stuff, but rather noted it's existence and noted any links?

So then a few years later, I need an install date and serial number for
some piece of equipment. Both those data points were on the equipment
invoice from the supplier, and the customer invoice, plus maybe some
notation on my calendar. I could search the invoices through my financial
software. I could scroll through, search a bit through my calendar to find
a date range. Then I could look for a supplier invoices (pdf, scanned
documents, etc.) around that time until I find what I need.

Or the indexing, with ontologies, can give me the documents tied to the
customer. Maybe the job. And narrow by supplier. And then some fine tuned
simple data indexing. And I get what I need.

This stuff can change how we write applications. It can make it useful and
organizable to run whatever application serves best for the specific
purpose you need, without having to worry about losing the link to a
defined workflow.

Indexing is useful, but tying it to structured data formats like these
ontologies makes it powerful.

Derek

I have approximately 3,500 songs on my laptop. There's only so much a folder structure can do with that many songs. Thankfully, most decent media players will index the music and I'll be able to search via tags. Nepomuk is an extension of that, opening the info to applications OTHER than just the media player that has indexed it (and, potentially simplifying the development of media players).

I also have over 2,300 pictures on this computer from various events and trips, each with different people. I have over 1,100 pictures from a single trip, taken in various cities. There are some photo management applications out there, like Digikam, but they're usually rather large and complicated. So far for that trip I've only "sorted" (i.e. putting them in named folders) about 200 of the pictures, and only by the city (most of which in a single city). Being able to also easily tag it with info such as what it specifically is of, people in it, the day, etc, would be VERY nice... it'll also be VERY nice to be able to easily find the picture based on the tags, even OUTSIDE such photo management applications.

Then, there's about a billion (no, not a literal amount, just more than I want to try to calculate) misc documents stored in a variety of locations, many created/downloaded on this computer, and many more recovered from a prior system that died on me. At first, the organization worked fine, but the number of documents (PDFs, word documents, spread sheets, power points, etc) has FAR out grown the folder structure.

A clever folder structure could help in all those situations (I have my music sorted: Artist/Album/Song), but all of those methods enforce one way of attempting to find something, but fail when approaching it any other way (what if I remember the song name, but forgot the artist/album? What if I want to look at all the songs by a specific artist? etc). Symlinks could help those situations, but would require a MASSIVE amount of work, and require predicting all the possible ways I'd want to filter before hand.

Desktop search helps with this, and Nepomuk is basically desktop search++... or more so, an extension to it (which may or may not make the analogy accurate depending on how you feel about C++ :P). Nepomuk involves using the data that Desktop search collects, in addition to the data the applications ALREADY have (but often just opt to lose/force the indexer to manually discover), more intelligently. It's definitely not an easy problem, but it has the potential to be a great benefit for DESKTOP users that have thousands and thousands of documents... a conservative estimate of the number of *personal* files I have on JUST THIS MACHINE is 8,500... I also have over 1,300 emails on my email account, but I use webmail for that as I've grown to hate all the email clients I've used... including Evolution, Thunderbird, Kontact/KMail, and many other OSS ones (they're either slow, crashy, huge resource hogs, cumbersome, or have horrible search mechanisms... or all the above).

Folders work well when you have a small number of files. But as you get more and more files, they begin to fail (extra pain when you have multiple flash drives with differing folder structures!).

Open Dolphin. Make sure the Information panel is visible (Press F11 if not, or find it in View->Panels.

Select a file. Notice there is a bunch of data on the file in the Information panel now, including a star rating widget.

Rate the file 5 stars (make sure it's a full 5 not 4.5)

Now in the Search... bar, enter 'rating:10', press Enter, nepomuk searches and shows the rated file.

Alternatively, set a tag 'kde' using the Add Tag widget, then search for it with 'hasTag:kde'.

If you just enter a bare string, it searches the file content indexed by Strigi, but indexing may not be enabled by default. You can turn it on with systemsettings, Advanced->Desktop Search->Enable Strigi Desktop File Indexer.

I've got svn (KDE 4.4 pre) installed here and the search widget has gotten a lot smarter so it suggests the attribute you may want to search for as you type eg tag, title, rating, and logical operators. There is also the beginnings of a graphical search builder where you can add search terms and operators. "Rating:2 and hasTag:demo" worked perfectly.

When I noticed that Strigi also indexes the exif tags from my pictures, I was surprised to find that i could search for Model:NIKON D90 and it worked. However, when I tried to combine that with 'and Rating>=4' the search service hung up, and it was unable to handle fields containing spacers like 'ISO Exposure'. I think it shows promise though.