NLP and its steps

Natural language processing (NLP) is the area of AI concerned with enabling computers to understand, interpret and generate human (natural) languages such as English or Nepali. It has two parts: natural language understanding (NLU), which maps text/speech to meaning, and natural language generation (NLG), which produces text from meaning.

Steps (levels) of NLP

 Input text/speech
       |
 [1 Phonetic/Phonological]  (speech only)
       v
 [2 Morphological analysis] words -> stems, affixes
       v
 [3 Syntactic analysis]     parse tree (grammar)
       v
 [4 Semantic analysis]      literal meaning
       v
 [5 Discourse integration]  link sentences
       v
 [6 Pragmatic analysis]     intended meaning in context
       v
   Meaning / action / response

1. Phonetic / phonological analysis: converts the speech signal into sounds (phonemes). Used only for spoken input.
2. Morphological (lexical) analysis: splits words into root and affixes and identifies word class. "Unhappiness" = un + happy + ness.
3. Syntactic analysis (parsing): checks that the word order follows grammar and builds a parse tree. "The cat sat on the mat" is accepted; "Mat the on sat cat" is rejected.
4. Semantic analysis: assigns a literal meaning to the structure and checks it makes sense. "Colourless green ideas sleep furiously" is syntactically fine but semantically odd.
5. Discourse integration: the meaning of a sentence depends on earlier ones. "Ram took the book. He read it." (he = Ram, it = book).
6. Pragmatic analysis: interprets meaning using context and intention. "Can you pass the salt?" is a request, not a question about ability.

Example: for "Ram eats an apple", morphology finds eat + s; syntax gives S → NP(Ram) VP(V(eats) NP(an apple)); semantics gives eat(Ram, apple).

Applications

Machine translation, speech recognition, question answering, chatbots and virtual assistants, information retrieval and extraction, text summarisation, spelling and grammar checking, sentiment analysis.

Why NLP is important for AI

• Language is the main medium of human intelligence; an AI must understand it to communicate naturally.
• It lets people use computers without programming, through speech or text.
• Much of the world's knowledge is stored as text, so NLP lets machines acquire and use it.
• It is needed for the Turing test, intelligent assistants, and knowledge-based systems.