Advanced Algorithms in Natural Language Processing
Spring 2017
Instructor: Dr. Yoav Goldberg Email: first.last @ gmail Office: 004 building 216 Office Hours: by appointment
In this seminar course we will discuss recent (and sometimes somewhat less recent) work
natural language processing and machine learning.
Students will read research papers and present them in class.
Course requirements
Each week, students are required to submit a review (details below) of the paper to
be discussed in class, at least 1 hour prior to class.
Each student is required to submit at least 6 reviews throught the course.
Each week, one or two students will present the paper that is being discussed.
The presentation should be about 15-20 minutes, and will be followed by an in-class
dicussion.
Each paper review should include a short summary of the paper followed by an opinionated
retivew on its conent.
Some questions you may use to guide your review are (many others are valid too):
Did you like the paper? Did you find it interesting? Be honest!
What are the most important things you learned from the paper? Why are they important?
Do the lessons learned generalize beyond the specific task? Do they promote our understanding of language? Do they contribute towards building an important system or application?
Is the experimental setup satisfying? Any experiments missing? Any obvious or important baseline missing? Is the ablation analysis sufficient?
If a theoretical analysis is included, do you find it satisfying? If none is included, is it missing?
Is the problem/approach well motivated?
Are you convinced by the results? Why?
Is the writing clear? Is the paper well structured?
Since this is not a real conference review, please also write what you learned form this paper and why, in your opinion, it was a good choice for reading (or why it was a bad choice).
Paper Presentation and Discussion Guidelines
Each meeting, if readings are discussed, one or two students will present the papers for 15-20 minutes. The presentation can use slides, can use the whiteboard, or can be just verbal.
When the paper revolves around a linguistic task, you should also follow the "data analysis guidelines" below. When a paper is primarily about a machine-learning technique, you should follow the "technical paper guidelines" below.
Data Analysis Guidelines
Pick at least 2-3 examples to discuss during your presentation in class. The examples may come from other sources than the paper itself. Pick the examples to illustrate various aspects of the paper and task. The questions you should think about include (but not limited to):
Why is this task difficult?
What are the hard cases?
What are the easy cases?
Can you think of a simple baseline? How well will it perform?
Why are the models discussed in class and readings appropriate?
Do these models make assumptions that hurt performance? How much do these assumptions hurt?
Is there an upper bound on performance?
What about the assumptions built into the annotation scheme? Any of them arbitrary?
If you see an example that is particularly fascinating, why is that?
Technical Paper Guidelines
What is the required technical background that is not in the paper and is required for understanding the method?
What are the main equations and/or algorithms in the paper?
What is the main innovation of the paper? How does it relate to previous work?
(Is the author's description of the previous work accurate? or misleading?)
What problem is the approach trying to solve?
Why is this a hard problem?
Can you think of a simpler baseline that will achieve simialr results?
Can you relate the approach and solution to specific data examples? (see "data analysis guidelines" above)
Is the approach specific to the current work, or can it be applied elsewhere?