Research Project Specification
Choose a topic and write a "thesis proposal." The topic
could be a research problem presented in class
or something different. Use the CS department's MS thesis
format, except that your outline should be for a journal
paper, and you do not need a signature page, a schedule,
a list of artifacts, or a list of delimitations.
Hand in your written proposal on the date specified
in the schedule. Proposals will be graded on the basis of
100 points, awarded as follows.
- 25 points for the introduction -- clear motivation for the
- 20 points for the
thesis statement -- clear, concise statement of the problem
being solved. (1 paragraph)
- 25 points for methods -- an explanation stating
how the problem is to be solved and the technical details
needed to make the solution viable. (2 pages)
- 15 points for the
contributions. (list 3-4)
- 15 points for the outline -- make this be a
two-level outline for a research paper you could publish in a
CS journal (in making your outline, consider how papers are
written for our discipline and why, and make your outline
conform to this standard).
Topic suggestion: Person Identification
Linking documents about the same person is an important task that should
Currently, search engines do not have the ability to identify which
documents give information about the same person in their list of
returned pages for a query about a person. For example, it would be
nice if a search engine (e.g. Google) could group all the
documents pertaining to a person
(e.g., Li Xu, or David Embley, or <your name>). How could we program
a search engine to solve this problem?
If you are interested in attacking this person identification problem,
you can use it as the topic for your research project.
Please feel free to select another topics related to data merging problems.
If you choose a different topic, be sure to clear it with your instructor
by Friday, June 14th.