• Question: What was the most interesting thing about your collected data

    Asked by Quentin Joyce to Sheila, Piyush, Natalia, Gary, Dimitar on 5 Nov 2018.
    • Photo: Dimitar Shterionov

      Dimitar Shterionov answered on 5 Nov 2018:

      Hi Quentin,

      In my work I use data in two or more languages to make a program to translate between these two languages. So, let’s say if I have enough data for English and Irish I can build a computer program that translates ‘Nice to meet you’ to ‘Deas bualadh leat’. But we need a lot of data otherwise the computer will not learn enough. And so my data is very large, some times including more than 10 000 000 (10 million) sentences. If you or I write them down it will be more than 200 000 pages and it will take us more than 12 years. And yet, the computer deals with it in just few hours or a couple of days.

Comments