Can you believe it! Photodeus.com (and all sites under it) were down for several days because of rats. They ate through the optic fibers that connect this server to the internet. It took the ISP quite many days to get it fixed, but we’re finally back online!

Just writing this as a reminder to myself. Today I upgraded the Photodeus.com server software from Tomcat version 5.5 + JDK 1.6 to Tomcat 6.0.18 + JDK 1.7.0 ea (early access), and compiled native the APR libaries to boost performance.

Fun.

Today, once again, I have been reading various, in my opinion truly interesting articles that I found via  random clicking on Digg.com and Slashdot.org. For example, I ended up reading about Bennett Haselton and his crusade against spammers, how he has been suing them and brought them into court. I even went as far as to read a detailed court transcript where his case finally was dismissed, which I think rightfully so, cause he didn’t put forward his case clearly enough… Too bad of course, another alleged spammer goes free.
Oh how I love court drama and the precise semantics that’s needed to state your case for a victory. It’s nothing new of course, it happens everywhere and everytime in many arguments. The ones who talk loud, fast and clear usually get heard. Unless the other one makes up a damn good lie. :D

Anyways, I was supposed to write about the fun I did today.

I took some photos and videos of my brother, doing some parkour and free running tricks. Mainly just some concept shots for an upcoming silly little movie that we’re gonna make one day in the near future. The idea to this movie started from a weird MSN chat the other day with a girl from England I don’t even know… I’ve probably had her on my contact list for a few months, but this is actually the first time we ever talked. I still need to edit the test clips and see if my ideas look any good on screen.

Then later today I went to my cousin and snapped photos of their new family member, a black dog they’ve named Scooby. Luckily it’s doesn’t look anything like Scooby Doo (I totally detest that cartoon btw). I’ll post some photos on Flickr later. Young dogs grow insanely fast and they’re cute only when they’re young. Young dogs tend to be quick on their legs too. Constantly moving around and sniffing on me or my camera. It was hard to take any good photos. I might take some photos in daylight later this week, it already had started to get a bit dusky when I took the first photos outside so I had to crank up the ISO rating on the camera and use the flash, which I don’t like.

Didn’t just read articles and play around with my camera. I also programmed a bit on my Popmundo diary analyzer, Snuggly. Found out that it didn’t work all that well for UK English only for US English, but fortunately the error was quickly solved. I’m still not at all satisfied with the graph output and looks though, and the Apache Commons Fileupload component messes up the UTF-8 characters for some reason. Not a priority right now, but I’ll do fixes and changes to them eventually.

Of course I did read even more articles, such as Tom Robinson’s attempt to recover censored text using Photoshop and JavaScript (some sorts of iterative fuzzy OCR solver). I’ve known for long that you can script Photoshop effects, and much more I suppose, with JavaScript, but I’ve never actually seen it in live action or even tried it myself because I didn’t know where to start or what to do with it. After seeing a flash video I can now say it looks friggin’ awesome!
A comment on his blog suggested improving his pattern matching algorithm with Markov chains, so I went checked out the corresponding article on Wikipedia. Seems like a really good idea, but would of course require a huge database in the right language, so probably hard to implement with a Photoshop script. Then I was remembering how I once at work used the Levenshtein distance to find simple typos or errors in our electronic HTML manual. The algorithm I created was simple enough, but yet accurate enough to help me solve the problem I had.

Later on I found myself reading about N-grams and fuzzy string searches, which I both find highly interesting topics. Naturally I ventured into lots of very scientific articles about spell checking, matrix math, Levenshtein automaton, Turing tests (which happens to be one of my favourite topics, I am both anticipating and dreading the day when conversational AI is good enough to talk with us in a “human way”), language detection, information entropy, text encryption, hard-drive partition encryption… and so on… Can’t even list all the stuff I read, cause the list would be almost endlessly long. I do truly love knowledge, esoteric and common stuff, and don’t think any information is useless. Too bad many interesting scientific concepts go beyond my comprehension: “Some language models built from n-grams are (n − 1)-order Markov models”. After a bit of reading I searched with Google for N-gram databases, and found out that none else than, of course Google released one in 2006! By searching around for applications using it, amongst results I found Chris Harrison’s bi-gram visualisations on his site. Pretty stuff! Be sure to click that link. Structuring and colouring words can actually make a nice poster.

Now, I’m really tired of writing this blog, I don’t even like to write all that much and often. Mostly because I can’t type as fast as I think. While I slowly punch my keyboard I’ve come up with a dozen new ideas to write about. Better that I go back to reading articles… and maybe I’ll blog more about my nerdy habits some other day. So long!:D

In the last few days I’ve been uploading photos to my flickr.com page (because my site is experiencing yet again some technical difficulties).

Go have a look!

This is for my friend in Holland, getting married today.