講師： Fogaras Daniel
(Budapest University of Technology and Economics, Hungary)
題目：Scalable similarity search algorithms over the webgraph
概要： Given a large repository of documents, similarity search
algorithmms enumerate the most similar documents to another document
submitted by a user. It is a challenging problem to serve such queries
on-line for a repository containing close to 8 billion web documents.
In my talk, I define the strict scalability requirements against
similarity search algorithms. Then, some similarity functions are
introduced that evalute similarities from the hyperlink structure
of pages (also referred to as the webgraph.) Finally, I will show
how Monte-Carlo algorithms can solve the scalability issue of
similarity search algorithms in some cases.