 |
Rice University
Department of Computer Science
presents
Panagiotis Tsaparas
University of Rome, "La Sapienza"
Link Analysis Ranking
Abstract
Ranking is an integral part of any
information retrieval system. In the case of the Web,
because of its huge size, the frequency of Web search,
it is even more critical. In recent years there has been a
surge of research activity in the area of Link Analysis Ranking,
where the underlying hypertext graph topology is used to evaluate
the relative authority of web pages. We will consider some of the previously
published algorithms, and will introduce some new alternatives.
Some of our algorithms can be described as non-linear dynamical
systems. Although the mathematics for studying
non-linear dynamical systems are not as well understood,
we will prove that the dynamical system we define converges
for any initialization, and we will study the combinatorial
properties of the stationary weights.
The increasing number of Link Analysis Ranking algorithms
creates the need for a formal way to compare and evaluate
them. We will define a formal framework for the theoretical
analysis of Link Analysis Ranking algorithms. Within this
framework we will formally define various properties
such as monotonicity, stability, and similarity.
We will study the properties of existing algorithms, and
we provide an axiomatic characterization
of the In-Degree algorithm (the In-Degree algorithm
ranks each page according to number of links that point to it).
Biographical Sketch
Panayiotis Tsaparas received his B.Sc. Degree from the Computer Science
Department of University of Crete in 1995. He continued his studies in the
Department of Computer Science of University of Toronto, under the
supervision
of Allan Borodin. He received his M.Sc. degree in 1997 for his work on
stability in Adversarial Queueing Theory. He recently completed his
Ph.D. degree
on the application of Link Analysis to Web searching and ranking. He
currently holds a post-doctoral position in University of Rome, "La
Sapienza".
His research interests include Models and Algorithms for the Web,
Information Retrieval,
and Data Mining algorithms.
Monday, Dec. 1 at 4:00 p.m. in DH 3076
Refreshments will be served
--- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- |
|
| |