CSE 291 Spring 2008

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING
UNIVERSITY OF CALIFORNIA, SAN DIEGO

CSE 291: Web-scale information retrieval and data mining

Spring 2008

Please use this discussion board to ask questions about CSE 291, especially questions about the projects.

CSE 291 is a graduate seminar devoted to recent research on methods from machine learning, information retrieval, and data mining that are useful for analyzing web-scale
datasets. If you want to learn about the algorithms that underlie many of the applications provided by Google, Yahoo, and their competitors, then you should take this course.

A partial list of specific topics to be covered is:

recommendation systems
collaborative filtering
probabilistic topic models for documents
Pagerank and newer alternatives
multilabel classifier learning
focused web crawling
sponsored search
search engine optimization
setting prices dynamically
optimizing the lifetime value of customers
online experimentation.

The course will include lectures and class meetings where a student gives a talk presenting a recent technical paper in detail. Class meetings will be on Tuesdays and Thursdays from 3:30pm to 4:50pm in the CSE building room 2154, beginning on Tuesday April 1.


date	topic or title	speaker	reading or handout	slides
April 1	Overview		list of possible papers
April 3	Improving the k-means algorithm	Sanjoy Dasgupta	www.cs.ucsd.edu/~dasgupta/291
April 8	Learning without negative examples	Charles Elkan
April 9	Wednesday 2pm: Special lecture	Yuri Lifshits	web algorithms
April 10	The k-medoids problem	Sanjoy Dasgupta	www.cs.ucsd.edu/~dasgupta/291
April 15	WebIC: A Complete Web Recommender System	Russ Greiner	http://www.web-ic.com/
April 17	Scalable Collaborative Filtering with Jointly Derived Neighborhood Interpolation Weights	Matt Rodriguez	http://public.research.att.com/~yehuda/pubs/BellKorIcdm07.pdf
April 22	Adversarial Classification	Aditya Menon	http://www.cs.washington.edu/homes/pedrod/papers/kdd04.pdf	slides
April 24	Google News Personalization: Scalable Online Collaborative Filtering	Jerry Fu	http://www.www2007.org/papers/paper570.pdf	slides
April 29	Automatic Identification of User Goals in Web Search	Emiran Curtmola	www2005.org/cdrom/docs/p391.pdf	slides
May 1	Semantic Annotation and Retrieval of Music and Sound Effects	Doug Turnbull	http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4432652
May 6	Clustering Billions of Images with Large Scale Nearest Neighbor Search	Dafna Bitton	http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4118757	slides
May 8	A Web-Based Kernel Function for Measuring the Similarity of Short Text Snippets	Jostein Oysad	http://www2006.org/programme/item.php?id=3069
May 13	On the Gravity Recommendation System	Matt Rodriguez	www.cs.uic.edu/~liub/KDD-cup-2007/proceedings/gravity-Tikk.pdf	slides
May 15	Estimation of the Click Volume by Large Scale Regression Analysis	Aditya Menon	http://yury.name/papers/lifshits2007click.pdf
May 20	PageRank for Product Image Search	Dafna Bitton	http://www.www2008.org/papers/fp506.html	slides
May 22	The Lane’s Gifts v. Google Report	Jostein Oysad	http://www.cis.upenn.edu/~mkearns/teaching/SponsoredSearch/Tuzhilin_Report.pdf	slides
May 23	The Economics of Internet Search	Hal Varian	http://www.calit2.net/events/popup.php?id=1328
May 27	Practical Guide to Controlled Experiments on the Web: Listen to Your Customers not to the HiPPO	Jerry Fu	http://exp-platform.com/Documents/GuideControlledExperiments.pdf
May 29
June 3
June 5

Each student who takes the course for four units will do a term project following specific guidelines. The project should be at the frontier of current research, and closely inspired by at least one of the papers discussed in the class. Many projects will be devoted to reproducing the results of a selected recent high-quality published paper. Doing this has very high educational value and provides an excellent starting point for subsequent high-quality original research.

Project reports will be evaluated using these grading criteria. There is a schedule for handing in a detailed project proposal, a draft project report, and then the final report. The seminar will have no final exam. Letter grades will be based mostly on the final project report, but the presentations, participation in class and in the web-based discussions, and the intermediate project deliverables are all important also.

The instructor is Charles Elkan (Professor), whose office is in the CSE building, room 4134. Feel free to send email to arrange an appointment, or telephone (858) 534-8897.

Note: The seminar runs in parallel with a data mining contest sponsored by Fair Isaac, with cash prizes.

SOURCES OF RELEVANT PAPERS

Papers will be recent technical articles, mainly from high-quality conferences and workshops. Students may choose articles from this list of relevant papers, in consultation with the instructor. Suggestions of other relevant papers are very welcome. These papers may be published in many venues. The following conferences and workshops have a high density of interesting papers.

Relevant papers can also be found through the 2006 University of Pennsylvania seminar on sponsored search, and in this list of papers written by people at Google. Yahoo has a similar list. Also see the website on web algorithms of Yuri Lifshits.

PRESENTATIONS

The schedule of presentations will be determined as early as possible. Students should choose a date first, and then agree with the instructor about a paper to present. To find ideas, students can look at this list of possible papers and contact the instructor. Each presentation will cover a single paper (or at most two), to ensure that it is explained and discussed in sufficient depth. If you want to change your presentation date, please arrange a swap with another student and notify the instructor at least two weeks in advance.

The procedure for each student presentation is as follows:

Two weeks in advance: Finish a draft of 40 to 80 slides that present clearly the work in the paper. Make an appointment with the instructor to discuss the draft slides.
At least one week in advance: Meet for about one hour to discuss improving the slides. Bring a printout of the slides that meets the following criteria: one full-size slide per printed page, single-sided or double-sided, pages numbered. Make slides attractive but simple. Do not leave excessively large margins, but do not put too much on a single slide.
It is likely that a second meeting with the instructor will be needed.
Day of presentation: Give a good presentation with confidence, enthusiasm, and clarity.
Less than one week afterwards: Make changes to the slides suggested by the class discussion, and email the slides in PDF to the instructor for publication on the 291 website. Try to make your PDF file less than two megabytes.

Each presentation should be prepared using Powerpoint or similar software, and should consist of at least 40 slides. You must write out all important equations, and copy all important diagrams, charts, and tables from the paper into your slides. Presentations will be evaluated, in a friendly way but with high standards, using a feedback form.

We have a web-based discussion area. For each paper, each student is expected to contribute at least one thoughtful message to the discussion, before the presentation. A message may ask an interesting question, point out a strength or weakness of the paper, or answer a question asked by someone else.

REGISTRATION

The prerequisite for this seminar is at least one graduate-level course (at UCSD or elsewhere) in machine learning or a related area. Appropriate courses at UCSD include CSE 250B (Principles of AI: Learning) and CSE 291 (Probabilistic Methods in AI and Machine Learning). Many other courses may be appropriate also. Anyone who is interested, regardless of experience, is welcome to contact the instructor to discuss participation.

Students may take the seminar for a letter grade for four units, or for one or two units, satisfactory/unsatisfactory (S/U):

For one unit, a student will present one research paper and participate in all class meetings.
For two units, a student will make two presentations.
Four units will require two presentations and a project.

For four units, a student should register for CSE 291, section id 620781 for a letter grade. For one or two units, a student should register for the instructor's CSE 293, section id 620800.

Most recently updated on May 27, 2008 by Charles Elkan, elkan@cs.ucsd.edu