Natural language translation – the present
Now, let's get back to LT. Remember, our goal is to create a human-like representation of natural language within computers. How do we do it? Well, let's take a bit of inspiration from the past.
John Rupert Firth was a very famous linguist, and this was one of his most popular quotes:
"You may understand a word by the company it keeps."Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â -John Rupert Firth, 1951
This principle was the inspiration behind enabling Neural Networks to understand natural language. If you can understand the words around a word, you can understand the word itself. What does this mean? Well, with NNs, you can understand words with semantic vector representations of those words. You do this by training neural networks to predict context based off of the word provided.
Take the following sentence:
Tanmay and Rob are writing a book on Watson, a Machine-Learning-as-a-Service platform.
Let...
