SIGIR'98 posters: Optimizing Recall/Precision scores in IR over the WWW
Optimizing Recall/Precision scores in IR over the WWW
Matthew
Montebello
Department of Computer Science,
Cardiff University,
Wales.
Abstract
The rapid growth of the World Wide Web (WWW) and the massive size of the
information corpus available for access symbolizes the wealth and
benefits of
this medium. At the same time this immense pool of information has
created an
information overflow which requires users to revert to techniques and
tools in
order to take advantage of such a resource and enhance the effectiveness
of
online information access. Search engines were created to assist users
to find
information by employing indexing techniques and suggest appropriate
alternatives to browse. These search engines have inefficiencies and
are not
focused enough to the needs of individual users and little has been done
to
ensure that the information presented is of a high recall and precision
standard. `Recall' measures how efficient the system is at retrieving
the
relevant documents from the WWW, while `precision' measures the
relevance of the
retrieved set of documents to the users' requirements.
We present our experiences with a system we developed to
optimize the recall/precision scores. We attempt to achieve this
objective
by employing a number of search engines and user profiling in tandem.
Namely, we attempt to optimize:
- recall by aggregating hits from major search engines and other
previously developed retrieval systems,
- precision by generating user profiles and predicting appropriate
and focused information to specific users.
Our system is able to easily and inexpensively accommodate future
generations of
web-based retrieval systems and technologies. Our contribution to the
IR field is that we were able to incorporate several desirable
characteristics
from different techniques to optimize and personalize WWW searching.
SIGIR'98
24-28 August 1998
Melbourne, Australia.
sigir98@cs.mu.oz.au.