Query Optimization with User Interaction

Speaker:Weicheng Wang
        Department of Computer Science and Engineering
        The Hong Kong University of Science and Technology

Title:  "Query Optimization with User Interaction"

Date:   Tuesday, 25 October 2022

Time:   10:00am - 11:00am HKT

Zoom link:
https://hkust.zoom.us/j/465698645?pwd=aVRaNWs2RHNFcXpnWGlkR05wTTk3UT09

Meeting ID:     465 698 645
Passcode:       20222023

Abstract:

Extracting interesting products/records efficiently and effectively from a
large database is a critical problem in multi-criteria decision-making. To
solve the problem, I have been studying a few operators by combining the
user interaction framework. In this talk, I will present two proposed
operators concentrating on different products/records and output
requirements. The first operator considers that the products/records are
only described by ordered numerical attributes and aims to find one of the
most favorite products/records. The second operator takes the
products/records that are described by ordered and unordered numerical
attributes into account. It studies how to interact with users to learn
their preferences and find their favorite products/records. Compared with
the state-of-the-art, both proposed operators achieve a vital breakthrough
in reducing user effort, which is required during the interaction process.

I will also overview two issues of the current query operators waiting to
be resolved. (1) The effects of different product/record attributes vary
on different user preferences. (2) There would be mistakes or
inconsistencies when a user provides feedback during the interaction. I
will show future research directions regarding these issues.


****************
Biograhy:

Mr. Weicheng Wang is now a Ph.D. student in the Department of Computer
Science and Engineering at HKUST. His research mainly focuses on data
analysis, with particular interests in query optimization and user
preference learning. He has published several papers in big data-related
top-tier conferences, including SIGMOD, VLDB, etc.