Mnemosyne learning data torrent

[Peter Bienstman]

Hi,

Thanks to the work of Tomasz Melcer, there is now a torrent of the collected
learning data of Mnemosyne:

http://www.legittorrents.info/index.php?page=torrent-
details&id=29b67f580ed97d6a10429697d5006c3eb47b2f18

Please post anything you find out from it here :-)

Cheers,

Peter

[HorCri]

Thank you thank you! I hope I can learn something by analysing this
batch of data.

HorCri

[mongrel]

To HorCri (or anyone else looking at this data):
Have you started working on this yet?
Would you mention your tools and procedures?
Thank you.

[Peter Bienstman]

Code to parse the logs is in the 2.x codebase, in the statistics_server
subdir.

Cheers,

Peter

--
------------------------------------------------
Peter Bienstman
Ghent University, Dept. of Information Technology
Sint-Pietersnieuwstraat 41, B-9000 Gent, Belgium
tel: +32 9 264 34 46, fax: +32 9 264 35 93
WWW: http://photonics.intec.UGent.be
email: Peter.B...@UGent.be
------------------------------------------------

[Karen]

Does anyone feel like explaining what a "learning data torrent" is?
Or is it one of those - if you have to ask, you'll never get it? Just
wondering,
Karen

> email: Peter.Bienst...@UGent.be
> ------------------------------------------------- Hide quoted text -
>
> - Show quoted text -

[querido]

1. "Torrent" is a system for storing and transporting very large
files, like DVD or bigger. A torrent (noun) is one of those files. A
torrent "client" is the program you install that downloads the file
for you, plus something else: everyone running a torrent client
participates in a worldwide distributed server; everyone who receives
a torrent also transmits to others (according to rules I don't know
anything about). So, you contribute to a distributed server, with your
compensation being the big files you can download. What you see, when
you try to download a very big file, is that you're receiving from
many different torrent clients around your area, or around the world,
simultaneously. Also, if you have a copy of data other people want,
and you put it in the designated place and label it as participating,
your computer will send bits of it as it has free time. (I think you
don't have to participate in that second part, but your system then
gets labeled as a sort of non-ideal node, or something.) www.bittorrent.com

2. The "learning data" referred to is the giant collection of all the
mnemosyne repetition and scoring data that's been uploaded so far.

The importance of this is: The author of mnemosyne is looking for
proof that one algorithm is better than another before accepting any
of the many changes to the algorithm that people like myself keep
suggesting. Lacking any proof, I keep changing mnemosyne's algorithm
for myself, back and forth, and I'm not sure that I've accomplished
anything *at all*. So, this is a big deal for flashcard-system
theorists.

The original post in this thread informs us that someone has collected
this mass of information into a torrent file, and has put it in his
torrent client for anyone who wants it. So, that's a nice favor to
those interested. (And the more people download it, the more copies
will be in the torrent system, and the faster subsequent downloads
will be. That's it.)

[Karen]

Whew, that was great. Thank you for the info. So much to learn about
info sharing, compiling, etc.

> > > - Show quoted text -- Hide quoted text -

[Alan Jern]

Hi all,

I'm a psychology graduate student and this data set looks really
interesting. However, it would be really helpful if there were some
documentation somewhere about what the columns in the database tables
mean. Or can you at least point me to a few key files in the source
that will enable me to decode things myself?

Thanks,
Alan

On Aug 15, 2:13 am, Peter Bienstman <Peter.Bienst...@ugent.be> wrote:
> Hi,
>
> Thanks to the work of Tomasz Melcer, there is now a torrent of the collected

> learningdataof Mnemosyne:

[Peter Bienstman]

On Wednesday 09 September 2009 08:12:18 pm Alan Jern wrote:
> Hi all,
>
> I'm a psychology graduate student and this data set looks really
> interesting. However, it would be really helpful if there were some
> documentation somewhere about what the columns in the database tables
> mean. Or can you at least point me to a few key files in the source
> that will enable me to decode things myself?

There is some info about what the fields mean here:

http://www.mnemosyne-proj.org/help/mnemosyne-xml-dtd.php

Also interesting reading is this:

http://www.supermemo.com/english/ol/sm2.htm

Feel free to ask if you have more questions!

Peter

[Alan Jern]

Thanks for your help!

Alan

[Alan Jern]

Actually, I have one follow-up question.

The way I understand this, each item is represented by a single row in
the database with fields for summary statistics like the total number
of acquisition and retention reps. The database does not contain,
however, data about individual training events, that is, something
like "on presentation 10 of item 2, user gave a grade of 4". Is this
correct? And if so, is there any way to extract information like this?

Alan

[Alan Jern]

Nevermind, it looks like there actually is a row for every event. Very
cool. If I get around to taking a close look at this data I will
definitely let you know what I find.