circle.ch weblog by Urs Gehrig
27. August 2007
25. August 2007
Behold's visual concept search is based on the technique called "automated image annotation". This technique calculates probabilities of concepts being relevant to images based on the pixel content of each image. The models for calculating concept probabilities are estimated using manually annotated training images. Behold implements the automated annotation model proposed in Alexei Yavlinsky's PhD thesis. http://www.beholdsearch.com
10. July 2007
12. December 2006
Code v2  updates the original work. It is not, as Lessig writes in the preface, a "new work." Written in part collectively, through a Wiki hosted by JotSpot, the aim of the update was to recast the argument in the current context, and to clarify the argument where necessary. http://codev2.cc
28. February 2006
Update: Have also a look at zenphoto.org that some combined with Lightbox [2,3,4].
17. November 2005
a Linux-based, full-color, full-screen laptop that will use innovative power (including wind-up) and will be able to do most everything except store huge amounts of data. This rugged laptop will be WiFi-enabled and have USB ports galore. Its current specifications are: 500MHz, 1GB, 1 Megapixel.So far quite impressive, if you keep $100 in mind. Nevertheless I would like to add the following;
06. October 2005
But anyway. QR codes are kind of twodimensional barcodes .
While conventional bar codes are capable of storing a maximum of approximately 20 digits, QR Code is capable of handling several dozen to several hundred times more information.20 digits are not enough to encode a typical blog title, a link, a date string and some copyright information. As an example, the QR coded string for this blog post is: "QR coded blog entries, 06.10.2005 00:23:26, http://circle.ch/blog/p1773.html, CC-by-sa_2.5". The QR code library behind my implementation is written by Y. Swetake .
If it makes sense, to link from a QR code to a blog post, is not yet plain clear. But I could imagine a simpler, maybe automated navigation for mobile phones:
Update: The ETH Zürich has a project running called "Visual Code Recognition for Camera-Equipped Mobile Phones" . They also provide so called Recognizer software for Symbian OS based mobile phones.
21. September 2005
OpenLaszlo is the leading open source platform for the development and delivery of rich Internet applications on the World Wide Web.Although I read the first time of it today, it looks promising, at least if I have a look at the demo applications  and the "Laszlo-in-10-minutes" introduction .
29. August 2005
The Term Extraction service provides a list of significant words or phrases extracted from a larger content.It can be implemented by making use of the Yahoo! API . One might then need to sort out the result set a bit and use them as "tags". Since this blog has no cache and the API only lets me have a limited number  of requests a day, I will not dive into it yet ;) For those using PHP, there is a so called yahoolib availabe.
28. August 2005
printf("<a href='/blog/p%s.html'I preferred to chose this solution from having e.g. a switch with ranges of posts/category to point to different style classes like size1, size2, size3 etc.
03. August 2005
14. June 2005
Magnus Bodin came up with this idea years ago. It is just fantastic, regardless of the "USELESS SITE OF THE YEAR 1998" award. From todays perspective it is not at all that useless, just remind what others do .
06. June 2005
29. May 2005
07. April 2005
05. April 2005
DNKA is a search tool for remote computers. It acts as a web server by interacting as a layer between Google Desktop Search (GDS) and the user. It allows other users to search, view and download your files, emails, chats and web history.Basically it is wonderful; you get a full search access to the indexed content of a remote computer. IMHO this application should be filed under "As long as You know what You are doing.". I tried it out in an intranet and got full access to the root drive C:\.
Then later on I found Baagle . The principle is similar to the above but the tools are different. Baagle is a set of Perl scripts wrapped around the search engine Swish-e. Baagle provides a standalone webserver and an indexer. But compared to GDS it is up to you to configure and restrict the indexing to the data directories of your desire.
If you like to figure out a similar solution with PHP have a look at the earlier mentioned nanoweb or nanoserv webserver built with PHP  and a Swish-e wrapper written in PHP . Once decided to go for Swish-e (on Windows) you need a couple of tools to handle various file mime types, among others pdftotext, ps2ascii, antiword and one that I found today to convert Excel files to text, called xlhtml. Just search.ch for them ;)
Well, now the only thing I could not find is a few lines of code to get a configurable search field to be integrated in the Windows task bar as we know it from GDS. Anyone up with a solution?
30. March 2005
Loïc d'Anterroches by the way has implemented a PHP Naive Bayesian Filter  that I could imagine would help me out the most with the trackback spam. Captacha is no solution there and I would rather prefer the "Bayesian Way" compared to semi-automatic solutions like moderation, blacklisting et.al. By the way, the script by Loïc did work right out of the box. ;)
21. March 2005
16. March 2005
Allenfalls kommt man zu denselben Suchergebnissen, wenn man die Option "&help=linkzoom[d|z]" auch ganz weggelässt.
15. March 2005