This schedule is tentative and subject to change as the semester progresses.
| Date | Lecture | Reading | Assignment | ||
|---|---|---|---|---|---|
| Introduction, N-grams | |||||
| Tu | 1/26 | 1 | What is NLP? | SLP2 ch. 1 | A0: The Basics |
| Th | 1/28 | 2 | Why do we need data?; Basic Text Processing (Regular Expressions, Normalization) | Working with text in Python 3, SLP3 2.0–2.3, NLTK Book 3.2–3.10 | |
| Tu | 2/2 | 3 | N-gram language models | SLP3 3.0–3.4.2; Goldwater probability review through section 4 | A0 due |
| Th | 2/4 | 4 | N-gram language models contd. | A1: Language Models | |
| Tu | 2/9 | 5 | Overview of Linguistics | Features English is missing - but most other languages have; SLP2 1.1, SLP3 11.1 | |
| Classification | |||||
| Th | 2/11 | 6 | Classification: naïve Bayes; Noisy Channel Model | SLP3 4.0–4.5 | |
| F | 2/12 | A1 due | |||
| Tu | 2/16 | 7 | Lexical semantics: senses, relations, classes | SLP3 18.0-18.3 | |
| Th | 2/18 | 8 | Linear models for classification: features & weights | SLP3 5.0–5.1 | A2: Perceptron |
| Tu | 2/23 | 9 | Linear models for classification: discriminative learning (perceptron, SVMs, MaxEnt) | Daumé The Perceptron: 4.0–4.3; SLP3 4.7–4.8, ch. 5 (Further readings are suggested in slides) | |
| Th | 2/25 | 10 | Linear models contd. | ||
| F | 2/26 | A2 due | |||
| Sequential Prediction | |||||
| Tu | 3/2 | 11 | Parts of speech; Review | SLP3 8.0–8.3 | |
| Th | 3/4 | 12 | POS tagging: HMMs | SLP3 8.4.0–8.4.3 Eisenstein Notes, 7.1: Part-of-speech tagging | A3: HMM |
| F | 3/5 | GRADED QUIZ 1 | (See study guide worksheet) | ||
| Tu | 3/9 | 13 | Algorithms for HMMs (mainly Viterbi); BONUS: Discriminative tagging with the structured perceptron | SLP3 8.4.4–8.4.6, Appendix A; BONUS: Eisenstein Notes, 7.5 (no need to read beyond structured perceptron); Neubig slides | |
| Th | 3/11 | 14 | Annotation; Universal POS annotation activity (tagset) | ||
| M | 3/15 | A3 due | |||
| Distributed Representations and Neural Networks | |||||
| Tu | 3/16 | 15 | Distributional representations and similarity | SLP3 ch. 6 | |
| Th | 3/18 | 16 | Deep learning and neural networks | SLP3 7.0–7.1, 7.3–7.5 | |
| Tu | 3/23 | 17 | Neural sequence modeling with RNNs (Jakob) | SLP3 9.1–9.4 | A4: LSTMs |
| Hierarchical Sentence Structure | |||||
| Th | 3/25 | 18 | English syntax, CFGs; forming project teams | SLP3 12.0–12.3, skim 12.4 | |
| F | 3/26 | P0: Suggest possible topics, form project teams | |||
| M-F | 3/29-4/2 | No class or office hours: Spring Break | |||
| Tu | 4/6 | 19 | Syntax contd. | ||
| Th | 4/8 | 20 | Syntax, CFGs contd. (P)CFG parsing: Parsing as search; CNF. | ||
| F | 4/9 | A4 due | |||
| Tu | 4/13 | 21 | (P)CFG parsing contd.: CKY walkthrough | SLP3 13.0–13.2, Appendix C C.0-C.4 | P1: 1-2 page proposal due |
| Th | 4/15 | 22 | (P)CFG parsing contd.: CKY pseudocode, PCFGs, lexicalization | A5: Syntax | |
| Th-F | 4/15-4/16 | P2: groups meet with instructor & TA | |||
| Tu | 4/20 | 23 | Dependency parsing (Shira) | SLP3 14.0-14.4, 14.6 | |
| Th | 4/22 | 24 | Semantic role labeling | SLP3 15.0, 19.1-19.6 | |
| F | 4/23 | GRADED QUIZ 2 | Study guide to appear | ||
| Translation and Sequence-to-Sequence | |||||
| Tu | 4/27 | 25 | Statistical machine translation | SLP3 11.2–11.9 | P3: Progress update, including lit review, due |
| Th | 4/29 | 26 | Statistical machine translation contd. | ||
| F | 4/30 | A5 due | |||
| Tu | 5/4 | 27 | Neural sequence-to-sequence models (Austin) | ||
| Th | 5/6 | 28 | Last day of class; Context in language processing + Wrap-up | ||
| W | 5/12 | VIRTUAL PROJECT POSTER SESSION (12:30-2:30pm) | |||
| W | 5/19 | PROJECT REPORT DUE | |||
| Visit the GUCL website for NLP talks this semester and beyond! | |||||