Challenges in NLP

  • Semantic Ambiguity: ambiguity in a word, phrase or sentence which has multiple interpretations.
    • Lexical Ambiguity: ambiguity in meaning(sense) of a word(or morpheme) when the word has multiple meanings. E.g. “book a room” and “Read a book”. it is handled using Word Sense Disambiguation (WSD) and Relationship Extraction techniques.
    • Scope ambiguity: caused by ambiguity in scope of a part of a sentence. E.g. in “Everyone has a home.” is there a single home for everyone or a single home for each person.
  • Grammatical Ambiguity(Syntactical Ambiguity): When ambiguity is in regards to syntactic structure of the sentence because the sentence could be parsed in many syntactical forms. E.g. “Good food and drink” creates the doubt whether “drink” possesses the attribute “good” or not.
    • Attachment ambiguity: caused by uncertainty of attaching a phrase or clause to a part of sentence. E.g. “The police shot the man with the gun.” arises the doubt in the ownership of the gun.
  • Pragmatic Ambiguity: ambiguity in inferred meaning from context which doesn’t match the semantics of the text. E.g. “Do you know what time is it?” depending on context implies being late.
  • Co-reference Ambiguity(Anaphoric ambiguity): Ambiguity in pronouns referring backwards. E.g. in “The wolf was hunting yesterday. it was horrific.” has the pronoun “it” as it’s anaphor, which may refer to any of antecedents(“The wolf”, act of hunting, or “yesterday”).