Cooperative Query Processing
with Semistructured Data
The growth of the Internet and other data repositories
outside traditional databases has forced rethinking
in several tenets in database research. In traditional,
databases, it is expected that all conforms to a certain
structure, which can be determined beforehand and
to which there are no exceptions. By contrast, data
in web pages is loosely organized. Semistructured
data is a new paradigm in database research which
studies collections of heterogeneous, irregularly
structured data.
Querying semi-structured data poses some particular
challenges. Since semistructured data is accessed
by many kinds of users, including non-expert users.,
the chances of formulating a query incorrectly are
higher than in traditional databases. The lack of
a completely regular structure also increases the
likelihood of making a mistake when writing a query.
Finally, query languages for semi-structured data
still share the notion of exact answer with more traditional
query languages. Thus, they are inflexible in that
they require exact matches between the query specification
and the data in the database, and are unable to point
out closely related information to the user. Therefore,
systems that manage semistructured data could benefit
from cooperative query processing techniques.
Here, we propose new methods to achieve cooperative
query answering (CQA) in the context of semistructured
data. The goal is to make the database systems behave
in a way that maximizes information exchange be devising
strategies in which the system does not merely respond
to queries, but tries to collaborate with the user.
Instead of literally answering a query, the system
tries to provide related data which may help the user
obtain the information she needs.
The overall goals of this project are :
1. To develop a general framework for CQA in semistructured
data environment to expand the traditional notion
of answer.
2. To develop particular techniques to capture more
knowledge on semistructured data. An extended answer
would answer questions like: what is to be considered
part of answer in case of partial knowledge? What
is to be considered close to a given object if we
need to enlarge an answer.
3. To develop an implementation of the framework
and the techniques
Resercher
Dr. Sanjay Madria
Dr. A. Badia, Univerisyt of Louisville, KY
|