CSE 628 Introduction to Natural Language Processing

Spring 2014

Mon /Fri 1:00pm - 2:20pm at [LGT ENGR 152]

Instructor:

Office Hours: Mon 2:30pm - 4:00pm

Course Description:

This course intends to provide a general introduction to Natural Language Processing (NLP), the study of computational systems to understand and/or generate human language. Approximately two thirds of the class will cover fundamental concepts and techniques in NLP, such as language models, statistical techniques for sequence tagging, parsing, information extraction, and information retrieval. The remaining one third of the class will introduce students to some of the more recent development in NLP research, such as the use of constraint optimization techniques, and ideas for data acquisition without expert annotation. Throughout the class, students will be exposed to some of the recent research papers that connect the concepts learned from the textbook to exciting research questions that are practically motivated.
Note: The material covered in this class will not overlap with the companion class [CSE 507 Computational Linguistics]; this class concerns subjects that are more computationally-oriented, while CSE 507 will touch more on the linquistic theories and semantics.
Textbook: Jurafsky and Martin, SPEECH and LANGUAGE PROCESSING: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition , Second Edition, McGraw Hill, 2008.

Prerequisites: Familiarity with either Artificial Intelligence or Machine Learning is strongly recommended, but not strictly necessary.

Tentative Grading:

- One must attend 3 sessions out of 3 in total. If you miss a session, you may compensate for it by attending a selected subset of department colloquium talks as will be announced via Blackboard. Everyone must actively participate in each session. You can either make a comment, or ask a question, or answer a question --- you must do at least one of the three in each session, but desirably much more than the minimum requirement. Submit the written critique at the beginning of each session as a print-out. It is recommended that you will bring a printed copy of the paper to the class. You will receive a penalty in your score if you walk in late for the paper discussion. You cannot submit your critique after the session. There will be no student paper presentation.

Late submission: Each student may adjust his/her homework deadline upto 7 days throughout the semester without a penalty. (not 7 days for each assignment, but 7 days cumulatively for the entire semester). After then, 10% of score will be subtracted each day. This rule does not apply to the critique submission which is due at the beginning of the class for the paper discussions. If you're late, then it is considered to be that you decided to skip the corresponding session. This policy is to encourage students to submit quality work, rather than poorly composed work in a hurry. For the purpose of counting late submission, fractional values will be rounded up - for instance, late submission by 1 hour is counted as late by 1 day. If there are situations where the application of this rule can be ambiguous, I have the right to apply the rule as I see appropriate. If you have a doubt, consult with me first before making your own assumption.

Assignments are posted to Blackboard

Tentative Syllabus: (subject to change depending on the students' backgrounds and interests)

Week	Topics	References
*--- Slides (provided below for convenience) correspond to roughly 70% of the actual class material. The rest will be delivered on the chalkboard. ---*
01 [Jan 27-31]	Introduction [slides] & Language Models [slides]	J&M Chapter 1 & 4
02 [Feb 03-07]	Language Models & Information Theory [slides ]	J&M Chapter 4
03 [Feb 10-14]	Text Categorization & Machine Learning Basics [slides]	J&M Chapter 5
04 [Feb 17-21]	Part-of-Speech & Sequence Tagging [slides]	J&M Chapter 5
05 [Feb 24-28]	Part-of-Speech & Sequence Tagging [slides] Paper Discussion I (Feb 28 FRI) --- keywords: sentiment analysis, deception detection, text categorization: [1] Thumbs up? Sentiment Classification using Machine Learning Techniques; Bo Pang; Lillian Lee; Shivakumar Vaithyanathan; EMNLP 2002 [2] Finding Deceptive Opinion Spam by Any Stretch of the Imagination; Myle Ott; Yejin Choi; Claire Cardie; Jeff Hancock; ACL 2011	J&M Chapter 6
06 [Mar 03-07]	Hidden Markov Models[slides]	J&M Chapter 6
07 [Mar 10-14]	Maximum Entropy Models & Conditional Random Fields [slides]	J&M Chapter 12
08 [Mar 17-21]	Spring Break!
09 [Mar 24-28]	Context Free Grammars [slides] & Parsing [slides] Paper Discussion II (Mar 24 MON) -- keywords: sequence tagging, information extraction, relation extraction, distant supervision, question answering: [1] Distant supervision for relation extraction without labeled data; Mike Mintz, Steven Bills, Rion Snow, Dan Jurafsky; ACL 2009 [2] Answer Extraction as Sequence Tagging with Tree Edit Distance; Xuchen Yao, Benjamin Van Durme, Chris Callison-Burch and Peter Clark; NAACL 2013	J&M Chapter 13
10 [Mar 31-04]	Statistical Parsing [slides]
11 [Apr 07-11]	Beyond CFG [slides] & Machine Translation [slides]	J&M Chapter 25
12 [Apr 14-18]	Machine Translation [slides] Project Update Presentation (Apr 14 MON)	J&M Chapter 22
13 [Apr 21-25]	Discourse [slides] Paper Discussion III (Apr 21 MON) --- keywords: conversation, textual entailment, paraphrasing, writing style, machine translation : [1] Towards Conversation Entailment: An Empirical Investigation; Chen Zhang; Joyce Y. Chai; EMNLP 2010 [2] Paraphrasing for Style; Wei Xu, Alan Ritter, William B. Dolan, Ralph Grishman, Colin Cherry; Coling 2012	J&M Chapter 22
14 [Apr 28-02]	Information Extraction [slides]
15 [May 05-08]	Hypothesis Testing [slides] Final Project Presentation I (May 08 FRI) Final Project Presentation II (May 12 MON)