Instructor	Administration	About the Course
Format	Schedule	Resources

Instructor:

• Office hours:  Tu/Th 9-10am
• Office: 2215 DCL
• Phone: (217) 244-2919
• E-mail: kcchang@cs.uiuc.edu
• Secretary:  Heather Hall, 2120 DCL, (217) 244-6241, hhall@cs.uiuc.edu

Administration:

• 8/28/03:
• Welcome to the class!
• Please note the location change-- We will meet in 3211 DCL (the same time) from now on.
• If you have signed up but decided to drop the class-- Please email the instructor to update the enrollment. Thanks! (Likewise, please also let me know if you want to sign up.)

---

About the Course

To learn to answer these "meta" questions, this course will study meta-research, or research about research, in the area of information access and data management. In particular, we will research about what research to pursue, a critical activity for any successful research efforts.

To establish a concrete technical focus, we will situate ourselves specifically in the area of supporting ranked top-k queries for data retrieval: As database systems are facing new challenges in non-traditional fuzzy retrieval scenarios, top-k or ranked queries are crucial for matching data by "soft" conditions, such as similarity, relevance, or preference: A multimedia database may rank objects by their "similarity'' to an example image. A text search engine orders documents by their "relevance'' to query terms. An e-commerce service may sort their products according to a user's "preference'' criteria to facilitate purchase decisions. For these applications, Boolean queries (e.g., as in SQL) can be too restrictive as they do not capture partial matching and result ranking-- which top-k queries specifically support.

Thus, this class will take all of us, the instructor and students together, to identify and create specific research agenda in the general area of top-k querying. We will research about what research have been done, and we will research about what research to undertake. Toward this end, this class consists of three stages:

• The Core: State of the Art

In depth, we will study the state of the art of top-k query processing developed in the database community.
• The Context: Broader Foundation

Broadly, we will survey the general concepts of ranking, querying, and retrieval, in databases and other (whatever) relevant disciplines, e.g., social choice, information retrieval, and collaborative filtering.
• The Creation: Future Agenda

As the ultimate goal of this class, we will identify and develop new research directions for top-k querying, by writing mini-proposals.

Prerequisites

• Required: CS311 (or equivalent)
• Recommended: CS411 (or equivalent)

Format

The CORE Stage:

For each such "core" paper (well, only 8 of them), students will read before class and send a Discussion Question Suggestion (DQS) to the presenter, by 9pm on the day before the class meeting. A DQS should be brief-- with no more than six sentences-- which suggests a discussion topic (1 sentence) and provides some initial thoughts  (5 more sentences at most). Hopefully, students will suggest DQ that is interesting and thought provoking. These DQS submissions will help to form the basis of our class discussion.

DQS Submission:

 
• Everyone (except the presenter): Email your DQS, in plain text directly typed in the message, to the presenter, by 9:00pm of the day before class. Use the subject "CS497KCC DQS dd/mm", where dd/mm is the date for that paper, e.g., "CS497KCC DQS 09/02" for the first paper. The presenter's NetID is noted on the class schedule, in the form of "netid@" (e.g., zhang2 for the 9/4 class)-- The email address should be netid@uiuc.edu.
• Presenter: Please collect (just cut and paste) all the submissions and compile them into a single list of DQS (with also the authors name) in one plain text message. Email this DQS list to Kevin by 10:00pm.

The CONTEXT Stage:

The CREATION Stage:

As the deliverable, the student will write a 5-10 page mini-proposal to define and justify the problem and speculate a solution. During the semester, we will discuss how to write a research proposal, and study some successful examples.

Meta-Research Fun Investigation:

Evaluation:

• (The Core) DQS and Presentation: 10%
• (The Context) Literature survey: 80%
• (The Creation) Future Agenda Discussion: 10%

---

Schedule

Most of the class readings are linked to their Electronic Edition (EE) where you can download their online copies, as provided by ACM Digital Library, SIGMOD Anthology, or the authors. Note that ACM DL can be freely accessible from the UIUC domain (see Resources).

Important Dates

• Context survey report:: December 3, 2003

University calendar:  Academic Year 2003-2004

Resources

Finding Papers Online:

• UIUC Grainger library (You can access ACM (click the link "ACM Full Text") and IEEE(click the link "IEEE Full Text") papers without ACM or IEEE account. )
• DBLP Bibliography Server
• Research Index