K-tree, NMF and INEX 2008

Today I gave a presentation within my research group at QUT. It discusses the submissions I made for the XML Mining track at INEX 2008. This required classifying documents based on previously known examples (classification). Another task required grouping similar documents together with no prior information other than the documents themselves (clustering). I also looked at different ways to measure cluster quality using negentropy and document link graphs. The K-tree algorithm is part of my research. This is the first time it was applied to document clustering. The results for the entire track should be out soon. I will also be giving the presentation at the QLD IEEE Computational Intelligence Symposium.

Unqualified should is evil

I was recently discussing the use of the word should with my research supervisor. I detest the word because I believe that is often used very ambiguously.

To quote myself,

"It [the word should] seems sly to me. In my mind it makes the statement "I really want you to do this but I don't have the guts to say it to you directly". I do not mind the use of should when it is qualified. i.e. "you should take a shower if you don't want to stink". This way the intent is clearly stated and there is no ambiguity. If I tell you that "you should take a shower", it could mean that I think you should shower because I am hand-wavy dictator that does things because that's how I feel on the day or I just hate people that don't shower 15 times a day. The statement "you must shower now" clearly demonstrates that you are being draconian instead of being underhanded and ambiguous."

The word should is often used when someone thinks that it is the proper thing to do. The word proper implies someone else's idea of correctness. I strongly believe that people must not force their beliefs onto other people because it restricts different points of view. This sentence might seem to contradict itself because I am trying to force this belief onto you. However, I am only trying to convince you. You are free to take it or leave it. English is very ambiguous and this leads to my suspicion that it is pointless to form an argument in it.

If we look at different meanings of the word should then we find out how ambiguous it can be.

Dictionary.com

Google

Wiktionary

Maybe I should be doing my research instead of pronouncing my hatred of the word should (contradiction and bad pun intended).

Smash

Oh my. I smashed my car today. Not good. Anyway, here are the pictures. It gets towed Monday.

342345351

Outside St John's

Here is the outside of the place I stayed while in Sydney over summer. Quite impressive.

363366

Just a little TV

Here is my brand spanking new 50" plasma TV. It is the Samsung PS50P91FDX. I have been happy with it so far. The video quality is awesome and it replaces my AV receiver as the audio and video switching device. It does HDMI, component, composite and vga where as the receiver only switched component. According to some website (I can't remember the URL) it is best to burn in your plasma at 50% contrast for the first 50 -100 hours. This is because the phosphorous is most prone to damage in the earlier hours of its use. I don't know if this is actually true or just yet another urban myth. I am giving it a go anyway.

317320323326329

I'm Beached as Bro

This video of a beached NZ whale is quite hilarious. Great accent impersonation. I couldn't have done better myself.

My Shared News

I am sharing my favorite news stories via Google Reader. This is one of the many nice features of the product. Hope you find them interesting.

WinSplit Revolution

If you use Windows and a large monitor WinSplit Revolution is a handy tool. It easily allows tiling of windows using the ctrl + alt + number cobination. It uses the fact the number pad is of somewhat similar shape to a screen. You can find more at the  programs website. This reminds me of using nvidia grid lines but it is far more user friendly. It is not as powerful as some tile based X11 window managers such as wmii and xmonad. Then again it is not a window manager.