NLP:
- Deals text and speech
Related disciplines:
- Linguistics. What makes an utterance correct or not.
- Psycholinguistics: experiments area. E.g. word introduced twice in a sentence, audio and how it is procesed.
- Philosophy: what does words mean.
- Comptational linguistics: takes the aforementioned and process them.
Motivations:
- Understand how humans communicate using language.
- Language is a distinguishing feature of humans (very social).
- Gossip replaces grooming (lol).
- Collective memory: with language.
- Some diferences are grounded in linguistics: there are two words for blue in Russian.
- Technological.
Difficult:
- New words ("ecotourist"), word senses ("literally", "gay").
- Highly ambiguous: crane (grúa, grulla in Spanish), bank.
- Multiple possible syntactics structures for word sequences.
- Complex: many grammar rules, and some not even exist.
- Partial knowledge: He put the computer on the chair. It was (wobbly | still hot).
History:
- 1940's: Turing decyphering the Germans.
- 1950's: semiconductors, Lisp is invented, Turing Test, Chomsky.
- 1960's: minicomputers, Eliza.
- 1970's: prolog, speech recognition.
- 1980's: Internet, declarative representations.
- 1990's: explosion of internet, formal/statistical approaches in NLP.
- 2000's: mass storage.
Other stuff:
- Question answering
- Natural Language Generation
- Spoken Dialogue Systems: people does not like talking with computers.