Slashdot Mirror
Google Admits to Using Sohu Database
prostoalex writes "A few days ago a Chinese company, Sohu.com, alleged Google improperly tapped its database for its Pinyin IME product, stirring controversy on whether two databases were similar just due to normal research process. Today Google admitted that its new product for Chinese market 'was built leveraging some non-Google database resources.' 'The dictionaries used with both software from Google and Sohu shared several common mistakes, where Chinese characters were matched with the wrong Pinyin equivalents. In addition, both dictionaries listed the names of engineers who had developed Sohu's Sogou Pinyin IME.'"
Leverage! Leverage!
Let no one else's work cut short your edge,
Against the truth you can surely hedge,
So don't cut short your edge,
But leverage, leverage, leverage!
(One man deserves the credit! One man deserves the blame!
And Sergei Brin Ivanovich Lobachevsky is his name!)
"In the future, Google invents a time machine that's used by a rogue employee to travel back in time to give Sohu this database. It's clear then that Sohu stole our database."
"Stolen from Apple Computer" (whole story)
When caught making a mistake, they admit it, work to resolve it, and move on? ...
I think there are a few other companies who could learn from that approach
"Do no evil"
should be changed to
"Do just a tiny bit of evil"
which at this rate will probably end up as
"All your web are belong to us"
Google is going to release a statement that stealing code/data is not evil in China, and Google must fit in local cultures and abide by local laws.
Seriously, this is just pathetic. I am appalled by the Google apologists on slashdot.
Chinese input is a well established market; Google Giant forces itself into the market with a product that is very similar to existing ones and offers no innovation. That is not evil enough? They did this by stealing data and who knows what from others. Mind you that the data is not publicly available, so Google must have committed certain crimes to obtain the data.
For those who don't see what's the big deal: the mapping from ASCII sequence to Chinese character/phrase is not trivial; actually it is what Chinese input is all about.
> They have not complied with Sohu's requests to date.
:-)
One of Sohu's demands was to remove it. They did that, even prior to the cease & desist deadline, per the article. It sounds like they'll have to compensate Sohu next, which isn't overly surprising. As for where they got it, perhaps someone sold it to them? We don't know, so I'll reserve judgment about whether it was acquired in an un-Google "evil" way until we hear the rest of the story.
> It's not the first time Google have taken a fairly liberal interpretation of someone else's copyright either.
As for the copyright stance, I honestly don't care. Yes, I dislike Microsoft's hypocrisy concerning copyright, but I don't really give a damn about imaginary property at this point in time, and I don't see Google out there telling people that copyright infringement is evil, wrong, Communist and anti-American.
Frankly, I'm more inclined to distribute my works with only one request: that you do not acknowledge my authorship in any way. Of course, almost the only way to enforce that is to post AC
So now american companies are pirating chinese software? Oh the irony! :)
Exactly. Reading 95% of the comments for this story and yesterday's story, everyone seems to think that this is about stealing code. This is about Google using the same data to train an algorithm. Both algorithms make the same mistakes because they were trained using the same data, which contained incorrectly labled information. It is whether or not this data was publicly available that is the issue.
For (a horribly contrived) example: Lets say that I write some hand writing recognition software using a neural-net. In order to train my software, I use a large database of handwriting samples that I have found on the web. However, the person that compiled this database made the mistake of labeling all of the sample images of the letter 'n' as the letter 'q', and all of the images of the letter 'q' are labeled as the letter 'n'. Person B comes along and uses the same data set to train a naïve-Bayes classifier. Guess what? Both algorithms will make the same mistakes when it comes to the letters 'n' and 'q'. Not because I stole code from Person B, but because we used the same training data.
I'm not defending Google at all here. If they stole the data from Sohu, they should get in trouble. Based on the fact that Google is in the web-mining business, I would guess that they just grabbed this data off of the net, and someone forgot to think about if they had the right to use it.
There are a lot of misundertstandings about how IME works and how Google copied non-public databases. So let me explain.
... in general, the same sequence could map to many different words (could be several dozens), and you usually need to select from them by choosing 1, 2, 3, ...(the input bar will display them from which you could choose, somtimes needing page-down). A native implementation of pinyin is thus very slow and cumbersome to use.
IME accepts keyboard input and converts it into certain language characters. There are many different input methods that decide how to generate Chinese characters by using English keyboards, and pinyin is one of them (and the most popular one).
pinyin is popular because it's simple and bears almost no learning curve. However, it suffers the problem of aliasing. For example, "shi" under pinyin will convert into "" "" ""
A good implementation uses following approaches:
1. adjust word location by how frequently it's used in the past. So most frequently used words are shift to the front, making selection much faster. Typically they should fit into the first page (no scrolling required).
2. allow partial input for common phrases. This inputs a whole phrase at once, each character only requiring the first English letters. It speeds up input significantly.
So the quality of the pinyin method depends heavily on how well the input could guess and prioritize the guesses, and thus the dictionary that is being used. And generating this dictionary (keeping it both contemporary and accurate) takes a lot of time.
The dictionary is typically distributed together with the input method (or it wouldn't work). You could obtain sohu's dictionary by just installing its input method, and Google has likely obtained it this way. However, I don't think it's in an open-standard format, so Google probably has done certain reverse-engineering to be able to actually use it in its own software.
"The internet is about the free exchange of other people's ideas!"
Am I part of the core demographic for Swedish Fish?
I've been thinking about this. Throwing the evilness of Google aside for a moment, why should someone be able to copyright a listing of the phonetic pronunciation of an alphabet?
Let's just imagine how I might create this list. I would have to hire people who spoke the Chinese. Then I would ask them to record the pronunciation of each character that they know. This is pretty easy because in Chinese each character has only one pronunciation (per dialect, anyway). There are about 3500 characters that you need to know in order to be literate. And all of these people would have learned these at school.
But how did they learn them? Well, they had a textbook and they memorized the list from the textbook.
Wait. I can't just memorize a list from one book and put it in another book. That's copyright infringement. In order for it not to be copyright infringement, I need to make sure that my sources all memorized the pronunciations from different sources. That's going to be difficult.
But let's say I do that. Now I have a list of the 3500 most common characters. And with that, I've probably got 99% of everything that's in a newspaper. But that's probably not good enough. I probably want a list
of say 60,000 characters. Otherwise it's pretty useless in a general sense. Uncommon characters are uncommon, but you *will* bump into the words over time.
So where do I find these characters? Can I hire some guy that knows them all? It would be very difficult. The best place to look is in a book. But wait... what am I going to do? Every time I find a character my people don't know, look it up in a book? Why don't I just copy it from the book in the first place? That's just copyright infringement again.
Really, the task of creating this list authoritatively without infringing copyright is monumental. Probably the *only* way to do it is with a community project where people just submit the pronunciations they know.
But if I'm going to have a community project like this, what the heck do I need copyright for? What am I protecting? If everyone is going to contribute, everyone should benefit.
So, personally, I don't think one should have copyright on this kind of material (same thing for spelling). It's just not in the public interest. This goes doubly so now that we have the internet and creating these kinds of projects is very inexpensive.
OK, I've gone on long enough... But one more rant. What's with this "do no evil" thing? Isn't that setting the bar a little low. If I told my parents that I'd work hard not to be evil, I think they'd be somewhat disappointed in me. If Google wanted to actually "do some good" rather than "do no evil", they could start a community project to collect this data and share it with the world.
Sigh... I guess we'll have to wait for some guy in his garage (but here's betting that someone has already started something).
No, actually, "gook" is a term that originated in the Korean war for Korean people. Because many of the soldiers who fought in the Korean war were officers in the Vietnam war, their racial slurs were adopted and modified by a new generation, leading to great confusion about the origins of the term.
The etymology of the word gook is interesting, because it may be one of the few racial slurs that originated with a people's term for themselves. In Korean, guk means "country" and by extension a country's people; when it is not modified (cf. waiguk, outside country, foreigner) it is understood to be Korea or its peoples. Speakers of Chinese will recognize the word as having sintic origin (gúo, country, and wàigúo, foreign country, respectively, in Mandarin).
The term was appropriated by the Americans during the Korean war and used as a racial slur for Korean people in general, which must have been confusing to the Koreans (imagine someone using "American" as a slur for Americans to get an idea). Then, in Vietnam, the old "Asians are all the same" mentality prompted GIs to extend its meaning (imagine "American" being a racial slur for all white people, for example -- yes, I know many Americans aren't white, it's not a perfect analogy, deal with it).