Weblog


Weblog

Perl Slurp-Eaze

25.11.2003, 16:55 · Martin · Permalink

Line by line processing is fine, but it isn't the only way to deal with reading files. The other common style is reading the entire file into a scalar or array, and that is commonly known as slurping. Now, slurping has somewhat of a poor reputation, and this article is an attempt at rehabilitating it.

"Where I've Been"

24.11.2003, 22:29 · Martin · Permalink

... my copy of "Where I've Been" arrived today, all the way from the USA to Nürnberg in a nice padded envelope. I'm listening to it now, and liking it. I must say I didn't know if I'd like it, I just thought I'd give it atry and help Scott raise some money on the way. Well it's a nice feeling to listen to a good CD I have bought and paid for AND actually did some good for in the same act! No major label can do that for me nowadays. Thanks, Scott!

Gold, gold, gold

24.11.2003, 16:20 · Martin · Permalink

it's Mine, mine, mine...
Now if only such a medal could buy me that g4 [i|power]book (all lower case, see?), *grin*
Comparisons are Feedster-ous
That's what I learned from Martin Spernau of Traumwind, who wins the Feedster Gold Medal (TM)

that reminds me

22.11.2003, 12:47 · Martin · Permalink

That my installation of Search::ContextGraphis currently showing only very poor performance accuarcy wise... I had already observed that it works better for posts with only a low word-cont and seemed to have a sharp drop-off for longer posts... but it also seems to decrease in accuracy with the size of the dataset...

Here's someone else who seems to have taken a look at it some time back: throwingbeans.org

Just for a test I staret up the old VectorSpace script I had done waaay back... it does have more 'relevant' hits... hmmm if only it was not so very slow... I can't sensibly run it on a regular basis. It takes more than 2 hours for each run to complete on my current set of 1500+ posts.

Just take this post as example. The VectorSpace engine found all the relevant posts about my PowerBook decision making... whereas the ContextGraph found only one relation.. which is not really relevant.

Working with Bayesian Categorizers

22.11.2003, 12:19 · Martin · Permalink

There's been some discussion in the blog world about using a Bayesian categorizer to enable a person to discriminate along various interest/non-interest axes. I took a run at this recently and, although my experiments haven't been wildly successful, I want to report them because I think the idea may have merit.

I see one major problem with his setup: each post can and should be able to belong to multiple categories... But it's a very interesting read

some further thoughts on his blog... (Jon Udell: Working with Bayesian categorizers)

What I think is the one most cripling thing about 'training classifiers' is exactly the training part. What works for spam filterning (there's only two categories good|bad) will fail in case of the user needing to define her multiple cats, and do so well. My approach (which isn't working very well either) tries to circumvent that problem altogether. I'm looking for qay to have posts 'cluster' together, not categorizing them in any specific way beforehand. MY system of Perl scripts (hopefully) simply finds 'similars' to the current piece of text, and let's the user decide (ad hoc) which relations are relevant (in the given moment)
More on that from the past...

[Power|i]Book confusion

19.11.2003, 13:07 · Martin · Permalink

After Scott pointed me towards the new G4 iBooks, I'm slightly confused... the prices for new iBooks are just about in the same region I'd have to spend for a used PowerBook G4... okay, iBooks come with 14'' TFT max, I'd like to have more, but still... Oh the choices... I was thinking this might be a nice example case for the Eigenpolls idea I linked a few days back

Well, Feedster on the rescue! Looks like I'm not the only one with that kind of question (har har, who'd have thought)

MacOPINION : Charles Moore | G4 iBook or Titanium PowerBook - Which Is The Best Choice? (For You) (explained a lot to me)

Simon is having a similar thread on his blog

Pathalizer: Visual website usage analysis

18.11.2003, 22:33 · Martin · Permalink

Pathalizer is a tool to visualize the paths most users take when browsing a website. This information can then be used to decide how to improve the navigation of the site, and which parts are most worth improving and keeping up to date.

In short, pathalizer generates a directed, weighed graph from an apache log (but it's easy to modify the first part of the process to use pathalizer for lists of other events).

Now this sounds very much like I was planning to do some time.. if only the output of 'dot' wheren't so unseemly

on Dreaming

18.11.2003, 22:13 · Martin · Permalink

while I'm at it: Something interesting happened/changed in the last weeks. I didn't often remember my dreams in the morning, only on special occasions. I credited that to changes in my sleep schedule (like going to bed early)
Now in the last week or so all of a sudden I remember almost each morning... And it is definetly not a change in sleep habits this time. (If anything I've had less sleep and of lesser quality)
There's one suspect I do have in this: I took up drinking earl grey a lot over my day (and also in the evenings), partly/mostly replacing my need for coffee/cafeine...

Hunting for Power

18.11.2003, 21:51 · Martin · Permalink

Hunting for a PowerBook, that is ;)
I decided it was time to get myself a little something for my new career as a freelance webtech... so I figured a laptop might come usefull. Now for about 1.2K EUR one can buy a real snappy P4 hingy with winXP and a DVD-R blah blah... So I figured 'why not go all the way?' and started looking for an comparable used Powerbook G4 on eBay... looks like I'll be switching in the near future... althought it will cost me a little more than a new Wintel notebook, I'm actualy looking forward to spending some real cash on a 'Tool of the Geeks'...

Eigenpolls

13.11.2003, 13:06 · Martin · Permalink

An Eigenpoll is a new type of poll design to extract knowledge from a community.

The problem with normal polls is that it find the most popular option. But if you are looking for a solution to a problem you don't want the most popular, what you want is the best solution to the problem.

The course is that when not everybody have a equal amount of knowledge about a problem, then the one man one vote principle don't work.
Eigenpolls try to solve this by letting the voters compare and rank the options, and then use a algorithm similar to Googles PageRank on the result.

SHriMP

10.11.2003, 22:03 · Martin · Permalink

The SHriMP visualization technique was originally designed to enhance how programmers understand programs. This research tool presents a nested graph view of a software architecture. Program source code and documentation are presented by embedding marked up text fragments within the nodes of the nested graph. Finer connections among these fragments are represented by a network that is navigated using a hypertext link-following metaphor. SHriMP combines this hypertext metaphor with animated panning and zooming motions over the nested graph to provide continuous orientation and contextual cues for the user.

via Gavin

this weekend

10.11.2003, 14:37 · Martin · Permalink

I died, was downloaded to another persons brain, shared his body for some time and finally was reborn, and thankfully everything went according to plan.

Oh, the joys of LiveRolePlaying in a futuristic setting, and the joys of being one of the people who write the plot ;)

So this weekend my character finally ended his existance as a holographic lifeform, and went from being a general nuissiance to being a biological one.

Oh, and on the way he helped make a psychotic character shizophrenic also. We parted as friends, just for the record, and he will be teaching me things about life as a biological entity... Let's see what will come of that...

multiple versions of IE on one machine!!!

07.11.2003, 11:00 · Martin · Permalink

this is awesome! Install multiple versions of IE on one windows boy. No multi-boot, no VMWare, no hands needed.
[via mezzoblue]

Traumwind presents

06.11.2003, 15:12 · Martin · Permalink

more or les proudly:
  • Happy2gether.de - Layout was not done by me, only the HTML/CSS/DHTML part of things. It does have a neat - if I may say so - CSS/DHTML Image Gallery feature.... those thumbnails are only a simple <UL> list, large images are displayed as background-image of the righthand DIV...

soundmosaic

05.11.2003, 15:54 · Martin · Permalink

Soundmosaic constructs an approximation of one sound out of small pieces of other sounds.

The soundmosaic algorithm is: Split the target file up into equal-sized segments, or "tiles". For each tile in the target file, find the closest match in the source files, and replace the target tile with the tile from the source files.

sorttable: Make all your tables sortable

04.11.2003, 18:55 · Martin · Permalink

While the web design community gradually moves away from using tables to lay out the structure of a page, tables really do have a vital use, their original use; they're for laying out tabular data.

unobstrusive DHTML, yu betcha!

CRM114 Based Usenet filter

04.11.2003, 11:42 · Martin · Permalink

With a training set of under 200 messages across 10 newsgroups, it is able to pick out the correct kind of job listings from jobs groups (as well as do a fine job of selecting geography, too), find things I might want in for-sale groups and generally work according to the theory.

The bayesian classifier spook resurfaces ;)

CRM114 Discriminator

04.11.2003, 11:37 · Martin · Permalink

CRM114 is a Controllable Regex Mutilator and Smart Filter, designed for easy creation of filters for things like incoming email redirection, spam filtering, system logs, or monitoring processes. Filtering rules can be either hard-coded (such as regexes), soft-coded (calculated at runtime or read from an external file or process), or learned dynamically by phrase matching (by SBPH hashing). This makes it possible to create very accurate filters with very little actual work. Accuracies over 99.9% are achievable.

I know I've had that one before, but it seems it's being actively developed, and has some new features since I last looked

AT&T naturalvoices.com demo

03.11.2003, 15:07 · Martin · Permalink

found Spidering Hacks example hacks Hack 86: Robot Karaoke (PDF Format)

weg vom Spielzeugprovider

03.11.2003, 13:25 · Martin · Permalink

der Schockwellenreiter macht endlich was nützliches aus seiner WG Miete

Daume drücken, daß es nicht irgendwo einen Wassereinbruch gibt ;)


alles Bild, Text und Tonmaterial ist © Martin Spernau, Verwendung und Reproduktion erfordert die Zustimmung des Authors

Martin Spernau
© 1994-2026


amazon.de Wunschliste

Facebook me!
Google+

powered by Traumtank