Translation brain machine celebrates 60th birthday
5 Jan 2014 Alan 0
The translation machine also known as the 701 computer was launched sixty years ago this week by researchers at IBM and Georgetown. The machine had successfully been able to translate many sentences into English from Russian and this led to the scientists claiming confidently that translation in the coming few years would be handled fully by machines.
After sixty years a review by MIT technology makes an extraordinarily comparable declaration that they can find the linear conversion that maps one to the other or that they can interpret one language to another. A group of Google engineers said that this was simple but actually it is not.
IBM recognised that even in the 1950’s, to translate just a single section needs two and a half times as many directions to a computer as compared to the time one will take to simulate the flight of a directed missile. It is also very uncertain that the Google scientists find anything to be easy about this method that depends on vector space mathematics. It is approved that there is a striking ease in machine statistical translation for instance the Google translate.
The more information one has the better chances of them getting an excellent translation in the end. However, what does one do when they do not have enough information or if Google does not have the information, as it is not part of the public and free web that one is supposed to get?
This therefore requires one to come up with a new idea just as Google has done. This new method is one that automatically makes phrase tables and dictionaries without any assistance from humans and it is one that is to be an accompaniment and not substitute their statistical approach. The new method utilises information mining in order to make a comparison of the structure of one language to another and later create dictionaries and phrase tables accordingly.
This therefore means that Google will not be required to depend wholly on documents that are only presented in two languages to be able to make its translation quality better as it will have this new technique together with the other techniques.