memoirs
http://kk77.blog19.fc2.com/
babbles + gibberish + cogitations + etceteraja
http://kk77.blog19.fc2.com/blog-entry-91.html
Google N-gram Over the years, like it or not, Google has gotten "more evil", meaning they have gotten more business oriented, profit driven, and investor conscious. I guess when your stock is trading over $700 a pop, you really need to. But hey, they still do good stuff on the whole.Recently, they made public their Japanese N-gram data. A huge corpus of Japanese sentences/words from the web conta
Over the years, like it or not, Google has gotten "more evil", meaning they have gotten more business oriented, profit driven, and investor conscious. I guess when your stock is trading over $700 a pop, you really need to. But hey, they still do good stuff on the whole.
Recently, they made public their Japanese N-gram data. A huge corpus of Japanese sentences/words from the web containing around 20 billion sentences and 255 billion words. That is a lot of data. The fact that they made this data available is actually not surprising since they made public their English N-gram data a year ago (similarly a corpus of mammoth proportions).
This corpus is personally intriguing to me, and to those interested in natural language processing and machine translation, I would assume it would be likewise. The English data is available for $150 and the just released J stuff is a bit more pricey around $200-$500. The data cannot be used for commercial use which limits certain things... but regardless, the data is something computational linguists like me will want to "play around with".
It isn't quite as known, but Google has been doing some hardcore R&D in recent years in the translation/localization field. Their aim seems pretty evident to me; to make it so that web pages written in a specific language can be machine translated (in a high degree) in other languages. Simply put, it's globalization and internationalization over the web.
In terms of machine translation, their approach is mainly statistical based as of now. I imagine they do use certain rules but overall their system is more raw brute force SBMT (Statistical Based Machine Translation). I have toyed with the new Google Translate and it isn't exactly a thing of beauty but I noticed it does seem to be more accurate than their previous system which was running Systran (which is good for MT between Latin based languages but not so much for Asian languages). Their MT system is gaining more credibility in the l10n world and a lot of it is a result of their massive corpus. ]]>
Tech2007-11-06T02:53:15+09:00k2FC2-BLOG
http://kk77.blog19.fc2.com/blog-entry-90.html
亀田家はもういいです。 亀田家、死ねばいい。と言ったらちょっとオーバーだけど、ホントボクシング界からはいなくなってくれないかと思うこのごろです。ウザイですこいつら。チャンネル変えるたびにDQN顔で知性がない口調で話しているヤツラしか出てこない。見てて、こっちまでおかしくなりそうだ。昨日の謝罪会見、あれは誰がどう見ても謝罪会見じゃないだろ。亀田大殻、オマエはたしかに乳臭え未成年だがよう、「プロ」のボクサーだろう?「申し訳ありま
ちなみに、マリオパーティ8、良い感じです。 アメリカではもう2ヶ月ぐらい前からすでに発売されていて、先月アメリカにいたときに友達の家で遊んだが、中々楽しい。このシリーズ、基本的には初代N64のマリオパーティからあまり変わっていないが、Wii のコントローラが英語で言う intuitive (直感的) なので、ミニゲームの操作がとにかく良い。おすすめです。と言っても、なぜか日本ではまだ発売されていないか。 ]]>
Babbles2007-07-21T14:41:56+09:00k2FC2-BLOG
http://kk77.blog19.fc2.com/blog-entry-87.html
J.R. Sakuragi - 桜木ジェイアール J.R. Henderson, former UCLA baller from the mid to late 90s is now J.R. Sakuragi, a naturalized Japanese citizen.WTF~ was my initial reaction when I heard of this. I actually kind of remember him playing ball in college. He wasn't spectacular, but he was a really really solid player. I guess it's kind of hard to stand out when you had teammates like Tyus Edney, Ed+Charles O'Bannon,
J.R. Henderson, former UCLA baller from the mid to late 90s is now J.R. Sakuragi, a naturalized Japanese citizen.
WTF~ was my initial reaction when I heard of this. I actually kind of remember him playing ball in college. He wasn't spectacular, but he was a really really solid player. I guess it's kind of hard to stand out when you had teammates like Tyus Edney, Ed+Charles O'Bannon, Toby Bailey, and last but not least, Baron Davis.
Henderson, errr, Sakuragi now, has apparently been in Japan since 2001, so that's 6 years. I'm not sure if he's married to a Japanese woman, but my thinking is that he is. If he weren't, he would need like 10 years of living+working experience in Japan to become naturalized. The process is not that easy, and you have to be proficient in the Japanese language and culture also. Then again, the process can be rather loose for special circumstances, which seems to be the case here.
The timing of this announcement is a bit controversial. He becomes naturalized and immediately joins the national team for the upcoming Olympic qualifiers... hmmm, shady if you ask me.
The funniest thing from this news is his newly given Japanese name. Sakuragi? 桜木? Like Sakuragi Hanamichi (桜木花道) from the manga "Slam Dunk"? Wow, you got to be K-I-D-D-I-N-G me here. It's a normal Japanese last name, but the "Slam Dunk" reference/relation is just too much. Maybe J.R. is into manga/anime or his Japanese wife is, who knows.
桜木の姓には「日本の人々の希望、明るさをもたらす桜の木のように」との思いを込めたという。
hahaha, LOL. 笑える。
The thing that makes things even funnier is, J.R. actually kind of plays like Sakuragi Hanamichi in real life. An athletic big man who runs well, boards like a demon, but has an iffy shot (in his UCLA days, he had a set shot, not a real jump shot).
I hope he grows out his hair a little and dyes it red. That would be pure awesomeness.
