MCA-209
Natural Language Processing
Getting a machine to handle human language — language models and n-grams, part-of-speech tagging, parsing and word senses, then translation, summarisation and chatbots.
- Semester
- Semester 3
- Credits
- 3
- Scheme
- 3 lectures a week
- Contact hours
- 40 across four units
- Category
- Core Elective - II
What the paper is
Natural Language Processing is built on one observation: language is ambiguous at every level, and each level needs its own machinery to resolve it. The paper works up those levels — words, syntax, semantics, discourse — and at each one introduces the statistical model that handles the ambiguity.
Unit II is the mathematical centre. N-gram models, parameter estimation, smoothing and hidden Markov models are what make part-of-speech tagging work, and the same HMM reasoning underpins a good deal of what follows.
The last two units are applied and tool-heavy: NLTK, Word2Vec, naive Bayes and maximum entropy classifiers, then deployment with Flask, information extraction, machine translation, summarisation and chatbots. This is the part of the paper most likely to turn into a project.
What to concentrate on
- Be able to compute bigram probabilities on a small corpus and apply add-one smoothing by hand. It is the most reliably examined numerical question and it also explains why smoothing is needed at all.
- Part-of-speech tagging with an HMM — the transition and emission matrices, and Viterbi on a short sentence — is the other standing question. Trace it once fully and the concept stops being abstract.
- For the tools half, concentrate on the pipeline rather than the library: tokenisation, stop word removal, stemming versus lemmatisation, vectorisation, classification. Naming the stages in order answers a surprising number of questions.
Syllabus, unit by unit
Summarised from the approved scheme. Hours are the contact hours the scheme allots, which is a fair guide to where the marks sit.
Unit I · Understanding language and language modelling
10 hours- NLP overview: definition, history and challenges, applications
- Language as a rule; artificial languages versus natural language
- Language modelling: grammar-based and statistical language models
- Branches of linguistics: statistical, psycho-, neuro-, computational and socio-linguistics
Unit II · N-gram models and tagging
10 hours- Simple n-gram models, estimating parameters and smoothing
- Evaluating language models
- Semantics, discourse and pragmatics
- Part-of-speech tagging and lexical syntax
- Hidden Markov models
Unit III · Parsing, semantics and NLP tools
10 hours- Syntactic parsing and semantic analysis
- Lexical semantics and word-sense disambiguation
- Introduction to NLTK; NLP with machine learning and deep learning
- Word2Vec models
- Language modelling with naive Bayes and maximum entropy classifiers, applied to document classification
Unit IV · Applications and deployment
10 hours- NLP model deployment with Flask
- Information extraction
- Machine translation
- Automatic text summarisation, auto-suggest and auto-complete, chatbots
How it is examined
- Question 1 is compulsory, covers the whole syllabus, and is made of objective or short-answer questions. It carries 12 marks.
- The rest of the paper is four units matching the four above. Each unit sets two questions and you attempt one, so there are four more answers at 12 marks each.
- Examiners are asked to map every question to a course outcome using Bloom's taxonomy, which is why questions are phrased as “apply”, “analyse” or “evaluate” rather than only “explain”.
The pass mark is 40 out of 100 in every paper.
Learning objectives
- Expertise in natural language processing and its applications
Assumed beforehand
- Python programming language
Paired in the scheme with MCA-265 Lab based on Core Elective II.
Books it is taught from
- TB1Jurafsky and Martin, “Speech and Language Processing”, Pearson, 2nd edition, 2013
- TB2Tanveer and Tiwari, “Natural Language Processing and Information Retrieval”, Oxford University Press, 5th edition, 2015
Notes for MCA-209
Lecture notes, question papers and assignments filed against MCA-209 are in the library. Everything there has been checked by a moderator before it appears, and an account is free — new ones are approved by hand, which is what keeps the shelf clean.