Example-based machine translation

Hey kiddo, have you ever used Google Translate to convert a sentence from one language to another? Example-based machine translation works in a similar way, but much smarter.

So, imagine you're a robot that needs to translate a sentence or a group of sentences from one language, let's say French, to another language, let's say English. The robot has a very big memory and has remembered hundreds of thousands of other sentences that have already been translated.

Now, when someone types in a new sentence in French, the robot compares it to all the sentences it has already translated before. It looks for a sentence that means something similar to the new sentence. Once the robot finds a matching sentence, it uses that previously translated sentence as an example to translate the new sentence.

For instance, if your robot buddy reads a French sentence, "J'aime manger du chocolat," it matches it to the sentence it has previously translated, "I like eating chocolate." After finding the appropriate translated sentence, it transforms it a bit to suit the French sentence you're trying to translate.

After the transformation, the robot translates the French sentence as "I love to eat chocolate." You can see that the robot used the example it remembered to translate the new sentence.

That's how short translation engines work. They use big databases with already translated examples to translate your piece of text into another language so that you do not have to do the hard work manually.