2024-08-28_lecture_2

CS 5034/6134 Natural Language Processing

Lecture 2: What is NLP?

  • Exploration of communication methods with computers.

    • Imagined Communication: How we envision interacting verbally with computers.

    • Actual Communication: The practical methods we use to interface with machines.

Understanding Natural Language Processing

Definition of NLP

  • NLP is the subfield that intersects linguistics, computer science, and artificial intelligence focused on computer-human language interaction.

  • Also known as natural language understanding or computational linguistics.

Applications of NLP

  • Virtual Assistants: Tools that help users by understanding and responding to queries.

  • Machine Translation: Translation of languages via AI systems.

    • New technologies improve inclusivity with the example of Google Translate incorporating 110 new languages.

    • Introduction of Zero-Shot Machine Translation allows models to translate without prior examples.

Question Answering Systems

  • Watson as an example of a successful question-answering machine, capable of interpreting natural language to provide responses.

Levels of Analysis in NLP

Types of Knowledge Required

  • Morphology: Studies the formation of words and their structures (e.g., prefixes and suffixes).

  • Syntax: Investigates the arrangement of words within sentences; critical for understanding sentence structure.

  • Semantics: Focuses on the meanings behind words and phrases.

  • Discourse: Looks into the coherence and context of sentences across larger texts.

  • Pragmatics: Examines language in practical contexts, such as intentions behind statements.

  • World Knowledge: Refers to common sense and factual knowledge necessary for comprehension.

Syntax: Structural Ambiguity

Case Studies

  • Examples illustrating sentence structure ambiguity:

    • "Time flies like an arrow."

    • "I saw the Rocky Mountains flying to Seattle."

  • Highlights issues of interpretation based on syntax alone without semantic context.

Ambiguity in Language

Types of Ambiguity

  • Structural Ambiguity: Different interpretations based on structure (e.g., can craft different meanings from similar arrangements).

  • Lexical Ambiguity: Confusion arising from words with multiple meanings (e.g., "bank").

  • Referential Ambiguity: Uncertainty who or what the references in sentences point to.

  • Phonological Ambiguity: Variability in sounds leading to different interpretations of phrases (e.g., "knight" vs. "night").

Handling Ambiguity in NLP

Statistical Models

  • Statistical methods employed to mitigate ambiguity in natural language processing tasks, requiring model training and estimation.

Conclusion

Current Status

  • Overview of advancements in NLP and challenges remaining in ambiguity across various levels of linguistic analysis.