NLP

NLP

Natural Language Processing (NLP) is an Artificial Intelligence (AI) undertaking. It aims to process sentences written in a language such as English so as to extract meaning from them and represent this meaning in a manner that a computer can understand. If you want to portray yourself as an AI practitioner, simply say that amongst all AI disciplines, you like NLP best and then go on to describe how you do it.

• How do you make a computer understand English?
  
  To this, you reply, “That’s what NLP is for. You break up the sentences into parts of speech, and then parse it just as you would any other string, of course, ellipsis is one major hurdle.” (‘Ellipsis’ means an instance where the context is not available.)


• What is ellipsis? Intimidate your audience here. “You don’t know what ellipsis is? Say you reply to a question about your name with ‘John Smith’. Here, you haven’t specified what John Smith is and that’s ellipsis.”


• Your audience will probably ask you at this point, “So that’s all there is to NLP? Parts of speech and parsing?”
  
  Since you don’t know too much yourself, you launch off into unfamiliar words— “No, no, there’s actually much more. You have got to consider things like failure of substitutivity, you have to learn to use the FOPC and HMMs, construct working grammars and much more.”


• What is FOPC and HMM?
  
  Remember to learn the full form—First Order Predicate Calculus. It sounds nice and AIish. Of course, you deliberately use the abbreviations in the first instance, so you can expand them later. HMM stands for ‘Hidden Markov Model’.


• Finish off with: “Like I said, it’s basically the manipulation of the parse tree. You construct the tree and do all sorts of things with it. I can direct you to a tutorial if you like.”
  
  

tree and do all sorts of

things with it. I can direct you

to a tutorial if you like.”