CIS4930 Natural Language Processing with Python (TA)

Special topics course introducing the essential concepts, principles, and techniques of natural language processing, with topics spanning information extraction, grammars, disambiguation, and system modeling, classification, and evaluation.

Instructor: Dr. Pete Dobbins

Term: Spring

Course Overview

A CISE special topics course introducing the essential concepts, principles, and techniques of Natural Language Processing (NLP). It covers:

  • Python for text processing: frequency distributions, conditional frequency distributions, file I/O, and regular expressions
  • Corpora, training versus testing, overfitting, and Zipf’s law
  • Bigrams, collocations, and n-gram language models
  • Statistical tests for collocations: t-test and chi-square test
  • Parts of speech and tagging
  • Supervised classification, feature extraction, naive Bayes classifiers, decision trees, and maximum entropy classifiers
  • Evaluation: confusion matrices, precision, recall, and cross validation
  • Finite state automata, chunking, and context-free grammars

Homework

The programming assignments I helped write and autograde:

  1. Implementing core text-processing functions from scratch, without external libraries, to understand how NLTK works under the hood
  2. Building a k-gram language model as a conditional frequency distribution and using it to generate text
  3. Writing a regular-expression tokenizer, extracting part-of-speech features, and identifying collocations with a chi-square test
  4. Scraping Shakespeare’s Roman tragedies from the MIT Shakespeare site, tagging lines by type, and analyzing their meter
  5. Training a classifier to identify whether lines of dialogue come from a Shakespeare comedy or tragedy
  6. Writing a context-free grammar in NLTK to parse source code into code, block comments, and inline comments

Prerequisites

  • COP3530 Data Structures and Algorithms

Textbook

  • Natural Language Processing with Python, 2nd edition, Bird, Klein, and Loper, available free at nltk.org/book

My Role

I served as an Undergraduate Teaching Assistant (UF uses the term Peer Mentor). I co-wrote the six programming assignments and the exam questions, built the test suites and a Canvas-integrated autograder that downloaded submissions, ran the tests, and posted grades and comments, and held office hours.