Joey's Diary

Okay, here is the first attestation proves that I've got my feet wet.

In Oct.10, 2006 (GMT +8 timezone), Westernizer was registered as a project in Sourceforge.net. Take a look at this, here's how I describe Westernizer project in Sourceforge's project page.

Westernizer is a web-publish digital magazine, it's a MPL licensed project and it privides materials about American TV shows, Internet culture and cool new stuff. Westernizer publishes in Chinese language only currently.

Now I'm working on the very first issue of Westernizer. It's expected to released by the end of the month. The making of a magazine is exciting yet tough, but actually it's not as tough as I thought it was, it just that there're so many things you need to learn. I've been contributing all my spare time on it these days, honestly speaking it makes me feel exhausted, you might not believe but I never went to bed earlier than 2:00AM these days.

Now this is my dream, and I'll do what it takes to make it come true. And without question, by making this dream come true, I'll have a lot of fun

Please check out Westernizer's project site:
http://sourceforge.net/projects/westernizer

BTW: the website is under construct, so I just use the project page as default website currently. The first relese is expected to leak out this month, so stay tune

It was completely accident that I found Google was interest in OCR. I know you might have the same response as what I had at first: What? Google? OCR? But if you can I'll tell you a not-that-long story about this OCR digging

It all began from a assignment my project manager gave me this morning. He gave me a huge address list (in paper form!) and asked me to transfer it into MS Office document by the end of the day. My first responce was very clear: there's no way I type this damn stupid address list character by character, I must find a useful OCR kit to help me out.

I was completely right. OCR was necessary, but which one? As you guys know, Chinese is a extremely complicated language in its written form, it is a what people call ideogram, it has more than 10000 characters which each one is composed by many different strokes. Okay, let's get things clear, there're ton of English OCR software out there but for Chinese language, there're very few candidate.

Fortunately few means there're still some. In the beginning I turned to freesoft field for help (there're no open source Chinese OCR exist, to the best of what I know), and found there was a tiny kit called Mini OCR. Unfortunately this software disappointed me a lot -- its succeed rate of recognise was probably less than 20%. The author was encourageable to release such kind of a complicated software for free of charge, but he really needs do more rewriting of the kernel of this software to make it work.

There came the plan B, the commercial software. There're some Chinese OCR manufacturers out there and Hanwang (literally means Chinese Language's King) was one of the most famous. Very unexpected, when I tried to find any useful download link for any of its product, it all turned out to be dead link. Hopefully there were some other name Chinese OCR manufacturers, and Qinghua TH-OCR was the one that I used before. I managed to find a web link to it latest version and installed it. TH-OCR worked just find, it successfully identified most of the characters of that address list, I was saved! Viva!

Now, here's how the story of digging Google & OCR comes. My custom is, when find something interesting and useful (for currently or future use), there's no way I just let it gi without doing some research. That was exactly what I did to this OCR subject. When I used Google to find the download link, I was aware that there's a article from blogspot looked like it was telling story abotu Google & OCR. I kept it in mind. When I done my work, I quickly turned to that page, and it turned out to be pretty much a official announcement of a open source OCR priject called Tesseract OCR.

This announcement stated that the Tesseract OCR was a commercial software found by HP more than ten years ago and used to be a major player in the OCR field at that moment, unfortunately it was abandoned years later when HP decided to quit the OCR business. After ten years of waiting, it was brought to life again by the support of University of Nevada in Las Vegas and (mainly) Google. This time Google decide to make it a open source project. Nowadays Tesseract OCR host its project pages on Sourceforge.net, and releases only in source code form. Needless to say, Tesseract OCR has a long way to go to catch up the major commercial kits on today's OCR markets, but anyway it's a very good start, and especially if you consider it's backed by such a gaint company Google.

And there's another article which is even more interesting. 'OCR & Google' analyses the purpose Google supports Tesseract OCR project, and it considers that it's because of the need of Google's online book project. This article defiitely worths a reading.


For more detail about the story of Google & OCR, please visit these pages:

Announcing Tesseract OCR (from Blogspot, by Luc Vincent, Uber Tech Lead)
http://google-code-updates.blogspot.com/2006/08/announcing-tesseract-ocr.html
OCR & Google (from Windows Live Space, by search-science)
http://search-science.spaces.live.com/Blog/cns!9D2D609139C8C9EF!3145.entry
Tesseract OCR Project's website
http://sourceforge.net/projects/tesseract-ocr

After a three-week's learning, after reading through two different books about Enron bankruptcy (one of them was the book 24 Days which was discussed on my previous post), and watching a documentary film named Enron: The Smartest Guys in the Room, now I feel that I'm finally ready to put some comment on this once America's biggest bankruptcy event.

When Enron was play trick, investment banks knew, traders knew, management knew, everyone but the employee knew about the truth. Everyone was just back off and busy making money from this sinking ship.

This can be consider as a lesson learnt that when your corp's about to IPO, think twice. Yes, IPO means a lot of cash, but after that, everything changes.

Sometimes it's a good idea to keep the company private. Don't go public until you're ready on every counts. Keep in mind that company means to build to last, not just a temporary ATM for management in just a couple of years.

I feel so sorry for those employee who lost their retirement fund and savings. They had nothing to lose when Enron backrupted while management'd already cashed millions of money from the stock market.


The Ship is sinking, but no one stops it.

This will happen over and over again as long as the existent financial system doesn't change, it may happen to your corp, it may happen to my corp, but which one exactly is the next one, you never know ......

It's the first time I've ever found a serious competition. God they're coming very fast! Fortunately they're just stepping up as well. Gosh, never thought this game (that's right, I treat it as a game, but it's a game that requires you to dedicate all your energy to it) is getting such breath-taking. Anyway, I've gotta move fast, as fast as I can.

Now I'm quite sure that I'll ignore my current job in the next whole month. I'm not speaking of quiting or something, but I think I definitely need some time to let me focus on the plan and make things done. I know that's not quite fair to my current project since they've been experiencing lacking of workforce these days, but there's nothing I can do.

I've stepped into this competition, there's no way to turn around. I don't wanna have such vision that five or six years later, I feel regret that I didn't move fast that some people else steal my originality and got themselves rich. No way! That's my originality, I'll try my best to protect myself and my benefit, I'll make this business work no matter how tough it's gonna be. I'm born to win, and I know it.

Time means everything now, there's no much time to waste, time killing is unacceptable. Move move move!