CIS530

CIS 430/530 Fall 2011

Computational Linguistics

Instructor
Ani Nenkova
Office: Levine 505
nenkova (AT) cis.upenn.edu
Office Hours: Wed, 4:30-5:30; Levine 505

Teaching Assistant
Alexander Shoulson
Office: Moore 103 (SIG Lab)
shoulson (AT) seas.upenn.edu
Office Hours: Thu, 4pm--5pm; Levine 612

Class Schedule:

Monday & Wednesday, 3:00PM to 4:30PM, Towne 313

Course Administrator:

Brittany Binler, 311 Levine, melema (AT) cis.upenn.edu

Course description:

This is an introductory course to computational linguistics, centered on the fundamental questions of how a machine can learn to analyze, understand and produce language. The topics covered include speech synthesis and recognition, syntactic parsing, semantic interpretation, discourse and pragmatic inference, and sentiment analysis. Students will get familiar with standard practical tools and resources for automatic linguistic analysis.

Prerequisites:

Undergraduate students should have completed CIS 121 before enrolling.

Textbooks:

[REQUIRED] Steven Bird, Ewan Klein, and Edward Loper, Natural Language Processing with Python --- Analyzing Text with the Natural Language Toolkit, O'Reilly Media, 2009. (Free Online)

[OPTIONAL] Daniel Jurafsky and James H. Martin Speech and language processing: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition, 2nd edition, Pearson Prentice Hall, 2008. (Available on Amazon)

[OPTIONAL] Chris Manning and Hinrich Shutze, Foundations of Statistical Natural Language Processing, MIT Press, 1999. (Free Online)

Various supplementary readings.

Grading:

4 Homework Assignments: 60%
- Each homework will involve implementation, design decisions and empirical evaluation.
  1. Implementation related to a specific language task. All assignments will be done in Python.
  2. Approach selection: In each homework assignment there will be a problem where a task is given, but details of how to solve the task will not be specified. Students will use knowledge acquired in lectures and general reasoning to make choices and explain the motivation behind these choices.
  3. Result analysis: Each homework will involve an empirical evaluation which will allow us to precisely quantify how well our systems work, as well as to compare the performance of different systems. We will do error analysis in order to identify where a system fails and point to possible ways to improve the system.
- NOTE: Each homework assignment will consist of 5 sub-problems. One of these will be different for students registered for CIS 430 and CIS 530. CIS 530 students will emphasize implementation, while those in CIS 430 will focus on analysis of system output. Over the four homework assignments this will amount to 10% difference in requirements.
Class Project: 35%
- The project can be implemented in any programming language.
- The project proposal will have to specify the project aim, the tools and resources necessary to achieve the aim, and the expected results and how performance will be evaluated. Students will get feedback from their peers and the instructor after their proposals. The goal is to define a non-trial task that can be reasonably accomplished by the end of the semester. Students enrolled in CIS 530 will be expected to complete a project that requires substantial implementation beyond the use of standard language processing tools. CIS 430 students will be encouraged to experiment with existing tools (part of speech taggers, language modeling toolkits, sentiment lexicons, syntactic parsers, etc) and integrate the outputs of several such systems in their class project system. 2 pages. (5%)
- [CIS 430 only] Presentation based on a scientific paper related to the project topic. The presentation will cover the aim, approach, results and their analysis of the selected paper. 15 minutes. (5%)
- [CIS 530 only] Related work write-up. Will have to discuss at least 6 papers related to the chosen class project. The write-up will compare and contrast the approaches described in this paper, identify weaknesses and opportunities for improvement. 4 pages. (5%)
- Final project write-up and code (25%)

Class Participation: 5%

Late-day policy:

Homework will be distributed in class and posted on the web page.

Late homeworks will be penalized based on the number of weekdays (or fractions thereof) passed since the HW was due:

1 day late: 20% penalty
2 days late: 30% penalty
3 days late: 50% penalty
>3 days late: No credit

For example, an assignment is due on Thursday at 4PM, if turned in late before 4PM Friday you will receive a 20% penalty, Monday before 4PM a 30% penalty, Tuesday before 4PM a 50% penalty, and you will receive no credit if turned in after 4PM Tuesday.

Academic Integrity:

Code of Academic Integrity

Back to Top

CLASS MODULES

General Information:: Using Python, NLTK, Coding Standards, etc.

CIS 430/530 Fall 2011

Computational Linguistics

CLASS MODULES

OTHER RESOURCES