CS6742 Fall : Nguyen-Vasilescu proposal

This page last modified Tue October 6, 2015 10:19 PM. Important updates will be posted to Piazza.

• The task (codename "Nguyen10Bias-Vasilescu14R"): By 1:30pm Monday the 5th (but earlier is better), individually post to Piazza (as a "Question", which makes it easier to track what hasn't been commented on yet) a roughly three-paragraph project proposal inspired by one of the readings listed for Lecture 13 (Tuesday Oct 6). In your proposal, include the general idea, at least one specific research question, and a suggestion for a dataset. If you have ideas along this front, mention what techniques/methods you'd like to try, as well.
• And, read each other's Piazza proposals, leaving suggestions as followups as you see fit, before class on the 6th.
• Things to think about:
  • What are interesting related problems?
  • What else could be done with the data, and/or what would be some interesting related datasets?
  • (If you have a good grasp of the techniques employed, then try the following. But otherwise, don't struggle through the details until you're sure it's worth it to do so. Life is short!) What else could be done with the techniques, and/or what else could be done to the techniques?
• Exception: the presenters for lecture 14 are exempted from this assignment, since they'll be preparing their presentations.
• No students are excepted this time: we are tentatively planning an activity other than proposal discussion for lecture 14.
• Assessment criteria: Proposals: thoughtfulness and creativity are most important to , but also take feasibility into account; Feedback: thoughtfulness and creativity of the feedback you give to others, on Piazza and/or in class (quantity is not a criterion per se, but you are expected to give thoughtful feedback to at least one proposal in at least one of Piazza and class).
• General notes on the propose-from-readings assignments:
  • The readings for this section of the course have been selected with several criteria in mind --- in no particular order, inspirational-ness, variety in topic and author lists, accessibility to those without an NLP background, availability or reproducibility of datasets; brevity; recency.
  • You do not have to read both papers, although we expect that in order to choose one of the papers, you might skim them both.
  • These assignments are not paper reviews, so don't include (irrelevant) paper summaries or criticisms. The intent is to use these papers as springboards for potential future work (yours or your classmates').