INFO60021 Natural Language Processing

1
INFO60021Natural Language Processing

• Harold Somers
• Professor of Language Engineering

2
What is NLP?

• Using computer to do linguistics
• Getting computer to handle language
• Synonyms (or components?)
• Computational Linguistics
• Language Engineering
• Basic tools, techniques and models
• Applications

3
What is NLP?
4
Some essential concepts

• Grammar as data
• Programs do something with the data
• Program uses the data to handle the input
• Declarative vs. procedural information (what
  vs. how)
• Formalisms
• If declarative independent of algorithm, thus
  reusable
• Algorithms
• Usually seen as searching a defined space for
  an answer
• Data structures

5
More essential concepts

• Analysis (parsing) vs. synthesis (generation)
• Using the grammar to certify input
• Using the grammar to produce output
• Transducer
• Specifies both input and output (defines a
  mapping)
• Associated algorithm has three uses
• Analysis
• Synthesis
• Verification

6
Even more essential concepts

• Resources
• Grammars
• Lexicons, word lists (WordNet)
• Secondary (re)sources
• Corpora
• Human-oriented dictionaries
• Empirical approaches
• Statistical, data-derived
• Machine learning

7
Syllabus

• Survey of applications
• Elements of linguistics, levels of linguistic
  processing
• Resources machine-readable dictionaries, corpora
  
• Computational morphology
• Finite state models
• Morphological analysis vs. Stemming
• Structural analysis
• N-grams, tagging, chunking
• Context-free parsing
• Probabilistic parsing
• Lexical relations
• WordNet, similarity in context
• Named-entity recognition

Assignment 1
Assignment 2
8
Assessment

• Examination 803 questions from 5
• Course work 20Two practical assignments, due
  weeks 5 and 8
• Will be based on grammar writing package which
  will be explained in class