|
TOPICS IN NATURAL LANGUAGE PROCESSING
|
Lecturer: Shay Cohen
Teaching Assistant: Nikos Papasarantopoulos
Time and Location: Monday 11am-12pm, 2.12 Appleton Tower; Thursday 11am-12pm, Lecture Theatre 2, 7 Bristo Square (semester 2)
Office Hours: Set an appointment.
Guidelines for course requirements:
(See also potential list of topics for fulfilling the course requirements here.)
Presentation
The goal of the presentation is two-fold: (a) to show that you have familiarised yourself with a paper or a topic so that you can actually freely talk about it, and answer questions from your peers; (b) to help you learn the skill of public talk: for some it may come naturally, others have to work harder on that. The more you expose yourself to this kind of experience, the better you will become. When choosing the papers for your presentation, keep the following guidelines in mind:
- You should choose papers or a topic so that the technical level of these papers match your technical knowledge. You are expected to be able to articulate the main ideas in the paper, and be able to interact with students who may have questions.
- It will help all other students if you prepare a handout that summarises the main ideas in the paper, similar to the ones I have been giving in class before every lecture. Such handouts can keep the students focused.
- This will not be mandatory, but if students agree, I will put their presentation online (and perhaps also the handout).
- Make sure you do not go over time. I expect presentations to be required to take at most 20 minutes with 5 minutes for questions. Practicing helps a lot with not going over time.
- Keep in mind the way the lectures were organised. Methodology first, followed by showing how they apply to different problems. Try to choose a topic where the methodology plays an important part, present the methodology, and then show what is the problem, and how the methodology was applied to that problem. (This is just a recommendation. I know from talking to you that some of you are more driven by specific problems they want to present. In that case, the methodology might take a more minor role.)
- Part of the requirement from the presentations is to come with one-two points for discussion in case there are no questions. The presenter will ask the audience these questions and look for response. This is part of the assessment.
Brief Paper Responses
The goal of writing the brief paper responses is to show that you can summarise an area or a paper in a succinct way, so that you could briefly explain it to someone else who is not knowledgable about the area. This is also an important skill to acquire, since it can foster collaborations with others, and enables you to communicate well about your research. When choosing the papers for your brief paper responses, keep the following guidelines in mind:
- It is required that you do the brief paper responses for papers that are chosen by other students for their presentations. If you send me the brief paper response before the presentations, I will put your summary online for other to read before the class, so they can better prepare. An ideal equilibirum for the whole class is such that each student generates brief paper responses for 2-3 papers belonging to student presentations.
- The above will not be strictly enforced, but as I said, it is recommended.
- You are also encouraged to post questions and comments about the different papers and presentations in the class forum before class, in order to have the presenter better prepare for questions students might ask, and in order to generate class discussion in general about the papers.
Essay
The goal of the essay is to show that you have deeply investigated a topic, studied the literature in the area and perhaps even ready to conduct research in that area. When choosing a topic, keep the following guidelines in mind:
- You should choose a topic that, on one hand, you feel comfortable studying in the relatively short time available, but on the other hand, challenges you and makes you learn new things.
- Your essay could be, for example, a literature review of a relatively significant number of papers in the area. Pick your papers carefully, try to make sure they are significant in the area you are focusing on (that is an important skill to learn as a researcher), and when in doubt consult with the instructor.
- You are more than welcome contributing your own ideas or work to the essay. Things you could do, for example: (a) describe a problem that you think needs a solution, but hasn't been previously addressed in the literature; (b) describe a possible extension or a path for future work following one or few of the papers; (c) describe experimental evaluation that you conducted that either replicates existing experiments in one of the papers or even a new set of experiments on a new dataset; (d) develop your own method and test it.
Here are examples of two excellent essays from 2015 (they are a bit on the longer side, your essay does not have to be that long):
Assignment
The goal of the assignment is to show that you were actively learning the material in the first part of the course, and that you were engaged in the lectures. This material should introduce you to the type of reasoning that is required
with modern NLP algorithm and model development.
The assignment is a set of pencil-and-paper exercises. The questions will be variations or extensions of the material studied in the lectures. You should be able to answer these questions when equipped with the lecture
handouts.
Back to main page.