I has been trying to build an English-Indonesian web translator. I had been searching for resources or algorithms I need to know. But from my experience it was so difficult to find.
What I need to know is what method do systems like "babelfish" use? Maybe some of you know more about this Machine Translation thing? Thanks for your help.
I'm planning to use phrase-by-phrase translation method. Is this method efficient enough? I'm planning to use it because I found that there is similarity on sentence construction between English and Indonesian.
Any comments?
i dont know about indonesian. but i think you can take two different approaches with different results:
- dictionary-like: just translate the sentence construction and give a dictionary like output for the nouns, verbs and adjectives. explain everything they can mean. then the user can take his brain to construct the sentence in his language.
- word by word: translate literally and everybody will have something to laugh...
you probably need a big database of groups of words that when used in the same context have a specific meaning. nothing for an after-work project though.
this probably was not much help, but i felt like giving my 2ec (euro-cent) to the problem in general...
Thanks M.Hirsch,
I'd like to explain what is in my thought. Let's say that I have a sentence, "I had bought the book".
Step 1: Get the phrases:
I
had bought
the book
Step 2: Translate each of the phrases
Step 3: Combine all the translation of the phrases in the correct order
What I'm trying to do here is extract the idea from each phrase and then combine them again to get one complete sentence again.
Is this kind of method efficient enough? Or, are there any better methods?
i donīt think this is effecient enough. but a itīs a good starting point.
i have no clue how to permutate words in english sentences, but there is probably several ways of saying something that are all right. in german eg, there is several ways to construct the same sentence.
another thing: you need to have CORRECT english for that. if someone doesnt care about punctuation (i seldom do when i just "chat" like here) it will be very hard to find the parts of the sentence.
like there is:
"if i had bought the book" - which has quite a different meaning but could not be translated by a script only using the phrases approach..
i am not good at grammar, but probably getting a university-level schoolbook about grammar could help a lot... and the help of a native english speaker that learned indonesian :)
..could not be translated by a script only using the phrases approach..
Hhhm, honestly, this is the biggest problem I have to face. I had realized it before. But I think it is beyond my abilities right now.
The problem(s) is about the ambiguities, in (word and sentence) meaning and sentence construction.
Hhhhm, it looks like I have to think more ...
Thanks Hirsch ...
We are discussing similar matters in another forum: forum.rtmt.info Someone asked about the advantages and disadvanatges of the different machine translation methods and I think we can share experience and cooperate. You can share your experience with us :)
We are discussing similar matters in another forum: forum.rtmt.info Someone asked about the advantages and disadvanatges of the different machine translation methods and I think we can share experience and cooperate. You can share your experience with us :)
What is wrong with me? I didn't know that it is wrong to ask or share experience. If this is your point, I am sorry to disturb you :confused: