NLP issues with example

Natural language processing (NLP) is the area of AI that lets computers understand and generate human language (text or speech) so people can communicate with computers naturally.

Issues (difficulties) in NLP

1. Lexical ambiguity: one word has several meanings or parts of speech. "Bank" (river bank / money bank), "book" (noun / verb).
2. Syntactic (structural) ambiguity: a sentence has more than one parse. "I saw the man with the telescope" (who has the telescope?). "Old men and women" (are the women old?).
3. Semantic ambiguity: the meaning is unclear even with a fixed parse. "The chicken is ready to eat."
4. Pragmatic ambiguity: meaning depends on context or speaker intent. "Can you open the door?" is a request. "Great weather!" in the rain is sarcasm.
5. Anaphora and reference: pronouns need antecedents. "Ram told Hari that he won" (he = Ram or Hari?).
6. Idioms, metaphors and figurative language: "It is raining cats and dogs."
7. Ill-formed input: spelling mistakes, grammar errors, slang, abbreviations, code mixing (Nepali written in Roman letters).
8. Ellipsis and incompleteness: "Ram likes tea and Hari coffee" (omitted verb).
9. World knowledge and common sense: "The city council refused the demonstrators a permit because they feared violence" needs knowledge to resolve "they".
10. Variation across languages and dialects: different word orders, scripts, rich morphology, few resources for languages like Nepali.
11. Speech issues: accents, noise, homophones ("their/there", "to/two").

Solving these needs lexicons, grammars, context, world knowledge and statistical/learning methods.