NLU and NLG; NLP steps

NLU (natural language understanding)

NLU is the part of NLP that converts human language input (text or speech) into a structured machine-usable meaning representation. It deals with ambiguity, context and intent. Example: from "Book a flight to Pokhara tomorrow" it extracts intent = book_flight, destination = Pokhara, date = tomorrow.

NLG (natural language generation)

NLG is the part of NLP that produces natural language text or speech from structured data or an internal meaning representation. Steps: content determination, sentence planning and surface realisation. Example: from weather data (temp = 28, rain = 70%) it writes "It will be warm today with a high chance of rain."

Steps of natural language processing

 Speech -> Phonetic -> Morphological -> Syntactic
        -> Semantic -> Discourse -> Pragmatic -> Meaning

1. Phonetic / phonological analysis: converts speech into phonemes and words (for spoken input only).
2. Morphological (lexical) analysis: breaks words into roots and affixes. "Walking" = walk + ing; "unhappy" = un + happy.
3. Syntactic analysis (parsing): checks grammar and builds a parse tree. "The dog chased the cat": S -> NP(The dog) VP(chased NP(the cat)).
4. Semantic analysis: finds the literal meaning: chase(dog, cat). It rejects "The idea chased the cat" as meaningless.
5. Discourse integration: links the sentence with the earlier ones. "Ram bought a car. He drove it home." (He = Ram, it = car).
6. Pragmatic analysis: gets the intended meaning in context. "Can you close the window?" is a polite request.

NLU covers steps 1-6 for analysing the input, and NLG performs the reverse path (meaning to sentence to text or speech) to produce the output.