CS 1671:
Human Language Technologies (Spring 2020)
|
|
Professor |
Dr. Diane Litman (5105 Sensq) |
Co-instructor |
Ravneet Singh (5501 Sensq) (ras306@pitt.edu) |
When & Where | Tuesday and Thursdays 1:00-2:15 (via Zoom) |
Office Hours | Litman: Th 2:15-3:30, by appointment; Singh: M,T,Th,F 10-11:30 (via Zoom) |
Description | This course provides an introduction to the field of natural
language processing - the creation of programs
that can understand, generate, and learn languages used by
humans. It will expose students to applications such as
chatbots by means of
computational techniques including search algorithms, dynamic
programming, hidden markov models, probalistic context free
grammars, and machine learning algorithms.
Prerequisites: CS 1501 (algorithms) OR consent of the instructor Textbook: Speech and Language Processing (3rd edition online draft - free!) |
Required Work | Homeworks (40%): written and programming Exams (40%): midterm and final Late Penalty: For assignments that may be accepted late, the penalty is 10% per day up to 5 days including Saturday, Sunday, and holidays. Assignments are due by 11:59pm. |
Date/Topic (tentative)
|
Textbook Readings
|
Assignments and Other Readings |
Jan 7, 9 Introduction (pdf) |
Ch 1 | xkcd Humor (credits to E. Riloff for this and links below) |
I. Words
|
||
Jan 9
Text Normalization (pdf) |
Ch 2 (2.1-2.4) |
regular-expressions.info
Humor: plurals, sentence tokenization |
January 10 (pdf),
14 [Singh] (pdf), 16 (pdf) Language Modeling with N-Grams |
Ch 3 (3.1-3.4) | Humor
HW1: assigned 1/16, due 2/4 |
January 21, 23 Part-of-Speech Tagging (pdf) |
Ch 8 (8.1-8.4.6) |
Humor
Schoolhouse Rock for Conjunctions |
II. Syntax
|
||
January 24 (pdf), 28 [Singh takes over for a while](pdf), 30 (pdf) Constituency Grammars |
Ch 12 (through 12.5) |
Do at home: come prepared to decode "Fish sleep" using Viterbi on 1/24 Humor |
January 30 (pdf), February 4 (pdf) Constituency Parsing |
Ch 13 | Humor
In class Grammar |
February 6 (pdf), 11 (pdf) Statistical Constituency Parsing |
Ch 14 (14.1-14.5, 14.8) |
Do at home: This is another example on going through the CKY algorithm that I made. The grammar is slightly simplified from the one shown in class. This is just an example you can use to review and practice the CKY algorithm.
HW2: assigned 2/11, due 2/27 |
III. Machine Learning
|
||
February 11 (pdf), 13 (pdf, Pictures of the Board), 18 (pdf) Naive Bayes Classification (and Sentiment) |
Ch 4 (through 4.8) |
Humor
Feb 13 Comment: Someone asked why 'F' was in the name of "F-measure". Going off of this paper here it was because of some name confusion with another function. Not a great explanation, but it is what it is. This paper also covers why we use harmonic mean and not normal average for F-measure. Feb 14 Extra lesson Materials posted on Coursweb in "Course Documents" section: Link to Courseweb |
February 20 (pdf) Logistic Regression |
Ch 5 (5.1-5.2; concepts in 5.3-5.7) |
9/26 NY Times: At Tech's Leading Edge, Worry about a Concentration of Power (evaluation of ML)
Midterm Review (pdf) |
February 25 (pdf), 27(pdf) Representation Learning (and Vector Semantics) |
Ch 6 | Word2Vec Tutorial |
March 3(Board Pictures) Neural Nets (and Language Models) |
Ch 7
Ch 9 |
|
March 5 | Midterm Exam (closed book) | Through 2/11 class (statistical parsing). No makeups. |
March 24 Dr. Litman returns and remote learning begins Zoom links and all future class lectures available via CourseWeb (today is NoahSmithPaper.pptx) |
Graded midterm review | HW3: assigned 3/24, due 4/2
From
word2vec to
BERT
(Bidirectional Encoder Representations from Transformers)
|
IV. Semantics
|
||
March 26, 31; April 2 Information Extraction |
Ch
18 (18.1-2) 3/31: Using New York Times Picks to Identify Constructive Comments |
Project: assigned 3/31, due 4/3 (hypotheses), 4/16 (complete project) |
April 2, 7, 9, 14 Word Senses and WordNet; Word Sense Disambiguation |
Ch 19 |
humor1,
humor2 |
April 14, 16 Semantic Role Labeling |
Ch 20 | |
April 16 (last class) |
|
Take Home Final Exam Assigned |
Saturday April 25 8-9:50AM (sorry, Pitt Exam Schedule) |
Final Exam Due |
Everything since midterm (i.e., not cumulative). |