subject: How to Translate Any Unknown Language! [print this page] Author: Said Benrida Author: Said Benrida
There are lots of translators that can translate texts from known languages in the world. But still these translators are not powerful enough and also cant translate texts from unknown languages; I mean really unknown ones like alien languages and very ancient ones. Then lets try to do now! The problem in actual translators is that they cant generate meaningfully correct output; they dont consider big differences between grammatical and conjugational rules. So they fail to give accurate translations. I thought about this problem and how can we also translate any language even if it is unknown. Then I found that we need two things: the right database and the right algorithm. We need first the most complete dictionary of the language which we will translate to, with all grammatical and conjugational rules and common examples and synonyms. We can add actual translators databases as they can help when translating known languages. Now we need a powerful processor with a high speed and a big memory to do the following tasks: Generate all possible translations of a text by replacing all the words figuring in the text by all the words in the dictionary database by following the following rules: Replace each word by one word. Then replace the first word by two word and others by one word each, then the first two words by two words and others by one each, then the first word by one word and the second word by two words and others by one word each, and so on until we try all combinations by replacing one word by one, two, three, until the number we desire of words, but we need to limit that number or the processor will not finish its work. Now we have all combinations and all possible translations, but most of them do not have a real meaning. To do we need to ask the processor to sort these combinations and select only the ones which are grammatically, conjugationally and meaningfully correct. To check the meaning the processor needs to browse the database following a certain order: The definition of the first word, then the definition of the words figuring in that definition and so on, then to move to the next word, and so on until finding a contradiction and then eliminate this translation, or close the process without any contradiction, and then add this translation to the translations list proposed to the user. Longer the text, less the number of proposed translations and more accurate are the proposed translations. Now what if we have many resulting translations with different subjects? We need to choose or to enter to the processor the context of the text or its main subject, and it will select the translation in direct relation with that context, or if not found, in indirect relation with it. We can use this technique to translate texts even for known languages as the computer can find the context and the related subjects in the database (the actual translators database). This technique still a theoretic one only and we need to test it. But if I am right then congratulations!About the Author:
Said Benrida
Web Marketer, SEO Manager and Traffic Manager since 2004.
