Skip to content

Latest commit

 

History

History
55 lines (47 loc) · 2.98 KB

project-reqs.md

File metadata and controls

55 lines (47 loc) · 2.98 KB
Raw

NLP1 17-18 Project Report

The report describes your project, the work that you have done and your findings. The report should be structured like a research paper.

Template

You are required to use an official template that is also used in research for the NAACL conference. You can download it from http://naacl2018.org/call_for_paper.html or directly edit it on Overleaf: https://www.overleaf.com/latex/templates/instructions-for-naacl-hlt-2018-proceedings/mbgbzpzxdkth

Length

Your report can be at most 8 pages long, excluding references.

Structure of report

For the report, we suggest that use the following structure:

  • Title, members of the group, student ids
  • Abstract (1 paragraph: the main idea and a key finding)
  • Introduction (max 2 pages):
    • Description of the problem area and the problem itself
    • What is the research question / goal?
    • Why is this an important / meaningful / interesting problem to consider?
    • The very basic idea of the approach and why this is a reasonable approach for this problem?
  • Problem: (roughly 1-2 pages)
    • Explain the problem; what kind of assumptions / observations you have about the problem
  • Approach (roughly 2-3 pages)
    • Explain the model; if any important assumptions are made at this stage, explain why they are reasonable or necessary
    • Explain learning / inference algorithms
    • Explaining (perhaps briefly) any necessary preprocessing / postprocesing / data acquisition stages (maybe earlier, depending on the project; may also move to the experimental section)
  • Experiments / Empirical evaluation (roughly 2-3 pages)
    • Any details about experiments (dataset sizes, parameter selection, etc)
    • Results
    • Analysis (discussion of results / visualization / findings / etc)
  • Discussion and Conclusions (0.5 – 1 page)
    • Refer to the research questions you defined in your introduction.
    • Any related work you are aware of?
    • Challenges you observed?
    • “Future work” (you do not need to do this work really J, but what would you change in the model / what experiments you would run / etc, if you would have a chance to do this? What other people should look into?
    • Any thoughts / observation / wider implications
  • Team responsibilities (1 paragraph)
    • Who did what

Write this report in such a way that it should be understandable to any student who took this class. In this class, we are not aiming for a very broad audience – this should be a paper accessible to a not-very-experience researcher in natural language processing or machine learning. However, make sure that the report describes interesting finding and your methods, rather than provides a detailed description of your program.

  • Do give a clear description on the decisions you made in the design
  • Use scientific writing style and language in your report (formal, clear, concise)
  • Use figures and tables to clearly display your results
  • Try to make the report interesting (i.e. concentrate on insights / general ideas / findings / etc – rather than on obvious details)