Query Personalization: when one size does not fit all
Georgia Koutrika
University of Athens
Talk
Georgia Koutrika has a PhD from the Dept of Informatics and
Telecommunications at the University of Athens.
She holds a M.Sc. in Advanced Information systems and a diploma in
Informatics from the Dept of Informatics
of the University of Athens. Her research interests focus on various
aspects of information access,
primarily in database systems, but also in information retrieval systems
and digital libraries, such as:
personalization, keyword searches, user modeling and user profiling,
query processing and optimization.
In this talk, we will challenge a traditional database (but also
information retrieval) assumption:
all users receive the same answer to the same query. In general,
however, different users have different needs,
interests, and background, which differentiate their expectations when
posing a query, making a one-size-fits-all answer
inappropriate to many users. This is particularly critical in the web
context, where the diversity of users is very high.
This weakness of current systems has been the starting point for the
work around query personalization.
Query personalization is the process of dynamically enhancing a given
query based on user-specific information
stored in a user profile with the purpose of providing a customized
answer. In this talk, we will describe
(a) a concise and expressive model for the representation and storage of
user preferences in profiles,
(b) algorithms for the selection of preferences from the user profile
that are related to a given query
and the generation of personalized answer based on the preferences
selected,
(c) the generalization of query personalization to a constrained
optimization problem, where constraints
may concern the execution cost of the personalized query, and the degree
of interest and size of the final answer.