Dear All:
Enclosed please find the Ph.D. Proposal Announcement of Ullas
Nambiar. Your presence is welcome!
regards
Rao
Computer Science & Engineering Department
PhD Dissertation Prospectus Defense
Answering Imprecise Queries over Autonomous Databases
by
Ullas Nambiar
Friday, May 14th, 2004
2:30 PM
BY 655
Committee
Dr. Subbarao Kambhampati (Chair)
Dr. Selcuk Candan
Dr. Huan Liu
Dr. Hasan Davulcu
Dr. Gautam Das (Microsoft Research Labs)
Abstract
Supporting imprecise queries over Web-accessible databases would allow
users to quickly find relevant answers. Current approaches for answering
queries with imprecise constraints require users to provide distance
metrics and importance measures for attributes of interest. Moreover they
assume the ability to modify the architecture of the underlying database.
Given that most Web databases are autonomous and may have users with
limited expertise over the associated domains, current approaches for
answering imprecise queries are not applicable to Web databases.
In this thesis, we propose a query processing framework that integrates
techniques from IR and database research to efficiently determine answers
for imprecise queries over autonomous databases having a relational
model. Specifically, we have developed two approaches for answering
an imprecise query by identifying and executing a set of precise queries
similar to the imprecise query. Our technical contributions include
(1) a domain-independent approach for deciding semantic distances between
values of categorical attributes, and (2) an approximate functional
dependency based approach for determining the relaxation order and the
importance weights of attributes. We demonstrate the utility of our
approaches and the accuracy of our answers by performing usability tests
and provide results.
Our proposed research will involve setting up a test bed to evaluate the
efficiency and robustness of our approaches over a fielded autonomous Web
database. Further, given that relevance is a subjective notion, we
will develop approaches to tune our results and similarity measures using
feedback given by the user. Finally, we plan to apply our techniques to
the problem of efficiently answering an imprecise query over multiple
autonomous distributed sources that overlap.
Open to public!