| Department of Computer Science and Engineering
University of California, San Diego |
CSE 250A
Winter 2001 |
For this assignment you may work in a team of two or three people.
The assignment has three stages, with two deadlines, February 18 and February
26.
(1) In the first stage, your team should find exactly six papers on a topic of your choice related to 250A. The set of papers you choose should satisfy all the following constraints:
(2) In the second stage, each member of the team should read each paper carefully and independently write a thoughtful review of each paper. Here is the template for reviews. As part of each review, each person should write his or her personal summary of the main contributions and findings of each paper. This summary should be an alternative abstract for the paper. When writing each summary, be careful to use your own words. Do not paraphrase text from the paper, and do not borrow phrases from it. In other words, you must avoid any hint of plagiarism.
(3) In the third stage, you should meet as a committee and discuss the six papers, your summaries. and your reviews. After you have reached a consensus opinion about the papers, if any original review disagrees markedly with the consensus, the author of the review should write a revised review that reflects his or her new final opinion. As a committee, you should also choose the best of the six papers.
On Monday February 26, you should submit all your revised reviews in
hardcopy, by sliding them under the door of the instructor's office.
Also submit a copy of the paper that you have chosen as best.
The suggestions below are useful also when you are looking for related work as part of a research project of your own, including for the other CSE 250A assignments.
You should use Citeseer, Cora, and/or the INSPEC database of technical papers available through Melvyl to identify your six papers. Citeseer is especially useful because it provides the full versions of papers, not just abstracts, and given one paper, Citeseer can list all the later papers that cite it.
If you choose to start with a specific topic and then look for papers, a good place to begin is the textbook by Russell and Norvig. For example, their Section 4.3 is about iterative deepening A* and related search algorithms. This would be an excellent topic to choose. The historical and bibliographical notes at the end of each chapter by Russell and Norvig are reliable and insightful.
Citeseer provides a list of the most cited papers in computer science. Papers from this list that are related to search, knowledge representation, and reasoning include:
Don't just choose papers that sound interesting individually. The papers you choose should have a common theme, and the theme should be reasonably central and important today. "Knowledge representation" is too broad to be a good common theme. "Approximating a knowledge base to achieve efficient question-answering" or "KR for semi-structured data" are examples of better, more focused themes on which one can find six papers that are related enough.
Papers are related sufficiently closely if you can compare them with each other directly for quality, and you can evaluate where one paper asks or answers questions that another paper fails to.
I also suggest that you avoid extremely mathematical papers. Mathematical results can be very interesting and important, but only if they address a problem that is already interesting and important. When reviewing a paper that is not purely mathematical, it is acceptable to write "This reviewer has not checked the correctness of the proofs in the paper."