|
|
|
|
![]() ![]() |
Jul 3 2006, 03:53 PM
Post
#11
|
|
|
Newbie [ Level 2 ] Group: Members Posts: 19 Joined: 3-July 06 Member No.: 14,279 |
Hmm, I guess Google's translations are better than Babelfish... but that still doesn't stop it from coming up with rather goofy translations on a regular basis.
I think the main reason that Google's translations aren't that great is that, of course, they're not being translated by a real live human being who has full comprehension of both languages. I'm pretty sure that language translation software will never be as good as a human translator. |
|
|
|
Jul 4 2006, 05:52 AM
Post
#12
|
|
|
Advanced Member Group: Members Posts: 114 Joined: 1-July 06 Member No.: 14,234 |
My experience with Google translation is limited to the translation on individual words when my mouse is pointing at a specific word. So far it seems pretty accurate surprisingly even in Chinese. But I’ve not tried using it to translate a whole sentence so don’t know if grammar will also be accurate.
|
|
|
|
Jul 7 2006, 12:53 AM
Post
#13
|
|
|
Premium Member Group: Members Posts: 200 Joined: 3-October 05 From: Missouri Member No.: 8,888 myCENTs:71.12 |
In August 2005, in a US government -run test, Google's translation application beat technology from IBM and from various universities. In the test, Google scored the highest amongst all competing software in Arabic-to-English and Chinese-to-English translation tests; these wereconducted by NIST (National Institute of Science and Technology). Each test comprised the task of translating hundreds of articles from Agence France Presse and the Chinese Xinhua News Agency from December 2004 to January 2005. How Google Wins? The answer lies in statistical analysis. It now seems increasingly likely that rather than have a system try to understand a piece of text, or formulate abstract representation taking context and other things into account, the most promising method involves looking at ready translations. How this works is that system would look at exisiting translations, and be trained on those. For example, if the German "reich" is often seen as being translated as "rich" in context of money, the system would pick it up. <snip> This is an improvement, but there is still a long way to go with translation systems. Most text simply cannot be accurately translated word-for-word. Worse, as translational systems get better, they will *look* like they are accurate but not be. Most language is not made of words but of rules, idioms and phrases. If you know the rules, idioms and phrases of a particular language, you can understand the text even if the actual words are encoded (look at cypher-solving problem like Treasure Island, etc.). Conversely, if you can translate the words 99% accurately but do not know the rules, idioms, and phrases, you cannot understand the text. As a simple example, Spanish allows double negatives: "No, no nunca voy" (I don't go ever, (I think)). Translating it word for word gives a worng or at least ambiguous meaning in English, and that is a simple example. Without translating more than one word at a time, you cannot do it right. In translating a web page or a news article, this may not be important. In academic papers (which is where I have needed it most), the ambiguity can be lethal (did that mean to mix in the HCL first always or never?). There were many times in college where you can not even get the sense of an article from its "translation". One good test of a translation system is to translate to a language and then back. You get some really funny translations, which is of course the source of the "invisible idiot" ("out of sight, out of mind") puzzles. |
|
|
|
![]() ![]() |
Similar Topics
|
Lo-Fi Version | Time is now: 5th December 2008 - 05:06 PM |