Project Research Project Due Project report due Monday

Project

Research Project • Due: – Project report due Monday midnight • Delayed a bit due to popular demand • Can accept an even higher delay (Tuesday at noon) – For a 10% drop in project grade. – Project presentation due to me and Tarique morning of class on Monday • 70% of 50% = 35% of your grade – 30% of 50% = 15% was decided by the mid-term report. – If you do really well, all mistakes are forgiven!

Project Report • At least 4 pages, at most 8 pages – Strict!!! points could be deducted if violated • (learn to be convincing not all-inclusive) – Includes EVERYTHING: references, figures. – Length is not evaluated! • Double column sig-alternate style – Use. cls and. tex From http: //www. acm. org/sigs/publications/proceedings-templates – If you don’t know La. Te. X, take this as an opportunity to learn (will help you in the future!). – Alternatively (but less preferable), use the Word format from here: • http: //www. acm. org/sigs/publications/pubform. doc

Project Report • START ASAP!! – If you’ve ever written a paper, you’ll know that writing the paper is almost as important as the research itself • Template (remember, 4— 8 pages): – 1 -2 page: Introduction – 0. 5 -1 page: Related Work • can be after Intro or before Conclusions – 1 -2 page: Problem Definition – 1 -2 page: Algorithms/Theorems/Models/Proofs – 1 -2 page: Evaluation + Experiments, Screenshots – 0. 25 -0. 5 page: Conclusions + Future Work

Project Report • Some of you have never written technical papers before, so I thought I should cover the basics so that you’re all on an equal footing – Also, some of you are used to writing other styles of papers, so making sure you understand the “systems/database” style. • Many Resources on how to write papers (if you are curious): – http: //cs. stanford. edu/people/widom/paper-writing. html – http: //web. engr. illinois. edu/~taoxie/publications/writepa pers. pdf – http: //web. engr. illinois. edu/~taoxie/advice. htm

Introduction • Think about catering to an EXTERNAL HARSH REVIEWER who does not understand/care much about your problem. You need to convince him/her that this is important and awesome. • Short, succinct, just enough: – Keep the reader interested • Did you cover the following points? – – – What is the problem? Why is it important? Why is it difficulty or challenging? Why hasn’t it been done before? Why do previous approaches fail? What is your novel twist? What are the key elements of your solution? • USE EXAMPLES!!!! – What is your list of contributions? • These 6 paragraphs should be present in EVERY paper you write. Take this as practice

Related Work • A good number of citations for a typical paper is between 15— 30. • Don’t just add citations to fill up the list – Actually identify relevant papers. – Google scholar is your friend. • For every paper you find, succinctly argue – What is it on? – Is it a different problem or the same problem? • If different problem can solutions be combined? • If same problem, why is yours better? • Aim to cover all bases: if you miss out on very important related work then you could lose points.

Problem Definitions or Setup • Be as formal as possible, while introducing as little “formalism” or “math” as necessary – Somewhat contradictory • Define all terms precisely (assuming an arbitrary database reviewer who may not have necessary background) – USE EXAMPLES!!!!

Core Techniques • Pseudocode is *not* useful • Explain the key idea with as little notation/math as needed – USE EXAMPLES!! • Provide proofs, arguments that it works • You should have the least amount of issues with this section (I think)

Evaluation • Especially important in a *systems/database* paper • Clearly specify: – The experimental setup • Goals of the experiment: – What are you studying? • Datasets • Algorithms/techniques implemented – (THINK ABOUT USING/IMPLEMENTING A BASELINE) • System/implementation details • Parameters varied • Then, each experiment – Each experiment should come with a clear “takeaway”

Conclusions and Future Work • One paragraph of what you have found, what are the avenues of future work from this point. – Future work is especially important here given that this was just a course project. – Tell me, if you were to convert this to a paper, what else would you have liked to have done?

Bonus • If you have a system demo: – Include screenshots – Host it somewhere (if easy) • Instructions on how to access the demo site – So that I can play with it

Project Report: Instructions & Summary • Midnight on Monday – Email title “CS 598 Final Report” – Tarique and me • Between 4— 8 pages • Include: – Introduction (remember the 5 -6 points) – Related Work – Problem Definition / Solution / Possible Demo – Thorough Evaluation – Conclusions and Future Work

Project Presentation • Monday before class – Email title “CS 598 Final Presentation” – 3 slides • You have 5 minutes to talk (I will use a timer!) • Essential to be present!!!!! – Extenuating circumstances: OK to send us a video before class. • Tell the class: – What you did – Why is it cool? • Why is it better than other stuff? • Why is it important? – Maybe a few sentences about the solution – If you have a demo, you can present that as well

Project Grading • 70% of the final project score still left: – 5% will be on your project presentation – 65% will be on your report • Of which 15% is on the writing • And 50% will be on the quality of the research • Usually the two are hard to tease apart, but will do something similar to the mid-term evaluation

Grades for Rest • We’ll send your – Paper reviewing scores – Presentation scores Later this week