Computer Laboratory

Course pages 2017–18

Overview of Natural Language Processing

Principal lecturer: Dr Helen Yannakoudakis
Additional lecturer: Prof Ann Copestake
Taken by: MPhil ACS, Part III
Code: L90
Hours: 18 (12 lectures and 3 x 2 hour practical sessions)
Class limit: 10 students
Prerequisites: No prerequisites beyond those topics covered in an undergraduate CS degree

Aims

This course introduces the fundamental techniques of natural language processing. It aims to explain the potential and the main limitations of these techniques. Some current research issues are introduced and some current and potential applications discussed and evaluated. Students will also be introduced to practical experimentation in natural language processing.

Syllabus

  • Introduction. Brief history of NLP research, some current applications, components of NLP systems.
  • Finite-state techniques. Inflectional and derivational morphology, finite-state automata in NLP, finite-state transducers.
  • Prediction and part-of-speech tagging. Corpora, simple N-grams, word prediction, stochastic tagging, evaluating system performance.
  • Context-free grammars and parsing. Generative grammar, context-free grammars, parsing with context-free grammars, weights and probabilities. Some limitations of context-free grammars.
  • Dependency structures. English as an outlier. Universal dependencies. Introduction to dependency parsing.
  • Compositional semantics. Logical representations. Compositional semantics and lambda calculus. Inference and robust entailment. Negation.
  • Lexical semantics. Semantic relations, WordNet, word senses.
  • Distributional semantics. Representing lexical meaning with distributions. Similarity metrics.
  • Distributional semantics and deep learning. Embeddings. Grounding. Multimodal systems and visual question answering.
  • Discourse processing. Anaphora resolution, summarization.
  • Language generation and regeneration. Generation and regeneration. Components of a generation system. Generation of referring expressions.
  • Recent NLP research. Some recent NLP research.
  • Practical on sentiment analysis. Students will build a sentiment analysis system which will be trained and evaluated on supplied data. The system will be built from existing components, but students will be expected to compare approaches and some programming will be required for this.

Objectives

On completion of this module, students should:

  • be able to discuss the current and likely future performance of several NLP applications;
  • be able to describe briefly a fundamental technique for processing language for several subtasks, such as morphological processing, parsing, word sense disambiguation etc.;
  • understand how these techniques draw on and relate to other areas of computer science;
  • understand the basic principles of designing and running an NLP experiment.

Coursework

Write a short report about a baseline sentiment analysis system based on the practical component of the course (up to 1,000 words).

Write a 4,000-word report including results from an extended sentiment analysis experiment.

Practical work

Build and evaluate a sentiment analysis system.

Assessment

Assessment will be based on the two practical reports. The first will be a ticked exercise, worth 10%. The second will be assessed by the lecturers and will account for 90% of the module marks.

Recommended reading

Jurafsky, D. and Martin, J. (2008). Speech and language processing. Prentice Hall (specific chapter references will be provided in the lecture notes).

Although the lectures don't assume any exposure to linguistics, the course will be easier to follow if students have some understanding of basic linguistic concepts. The following may be useful for this: The Internet Grammar of English