Project Idea
As discussed in class, your project can involve any of a variety of
themes. In particular, you may build an application, develop a
platform, perform data analysis, or conduct a
simulation–anything that demonstrates a connection to natural
language processing, involves nontrivial effort, and enhances (your
and others') knowledge.
I encourage brevity, hence the word limits for several of the
writing tasks. In a single-space format, a page is about 400
words.
Project Report R0 (10 points)
As the next deliverable (R1), I ask for a formal project
proposal. The following description is a precursor of what the next
deliverable will be (I will revise it based on the work you submit).
In the present deliverable (R0), please just approximate R1 as much
as possible, so your R1 effort is simplified.
In particular, at the outset (for R0b) you can begin with a half or
full page description of your idea, covering as much of the
following as you can.
For R0b, it is fine to submit two ideas and I can advise you on
which is better for this course. Subsequently, please submit
exactly one idea and deliverables for that idea.
I won't assign grades to R0 but submitting it is mandatory. Those
who submit R0 will earn the same points on R0 as on R1.
Project Report R1 (10 points)
- Write a two-page report with the following details. You can use a
picture if you like.
- Describe at a high level, clearly and concisely, the overall
problem you seek to address.
- If you build an application or platform, what
are its intended usage scenarios? Who will care to use what you
create and why?
- Describe any special features of your application or
platform. For example, if dialog context is important, then
elaborate on what features of dialog you employ (or exploit)? Why
does the problem naturally fit with dialog—who would be the
users and beneficiaries of your approach (members of the public;
enterprise managers; plumbers) and how will your proposed solution
reduce user effort, enhance usability, preserve confidentiality,
or what else?
- If you perform data analysis, explain the kinds of facts you hope
to discover or the hypotheses you will try to verify, what makes
those facts or hypotheses interesting, what datasets you will use,
how you will prepare the datasets, and what kinds of analyses you
will carry out.
- If you will conduct a simulation, explain what hypotheses you
will try to verify, what makes those hypotheses interesting, and how
would your simulation provide the information needed to prove or
disprove those hypotheses.
- Be creative. Don't overly worry about implementation details at this
stage.
Project Report R2 (10 points)
- Include a revised version of R1.
- Using up to an additional three pages, enhance R1 with a
description of the design of how your implementation is proceeding.
Concisely specify the main representations and reasoning you expect
to develop, explaining the key aspects of your design along with any
rationales. Include any preliminary, tentative results.
Project Report R3 (40 points)
- Include a revised version of R2.
- In up to one page, describe your implementation and any special
technical challenges you faced in bringing it to completion.
- In up to two pages, describe your results and other findings.
Describe what the reader should take away from your exercise.
Term Paper, due with Report R3 (100 points,
separate from the project, and to be scaled as indicated on the main
course syllabus)
- In up to two additional pages beyond R3, reflect upon your
project work with respect to
- The limitations of your problem formulation, hypotheses, methods,
and findings
- The prerequisites for taking your findings and applying them in
practice
- If you are preparing a formal paper for publication that addresses
the above points, you can submit that paper instead provided you
highlight or otherwise identify the sections where it addresses
these points.