Title: Recipes for graph mining - pagerank, communities and more
Speaker: Christos Faloutsos


Abstract:
First we shall give a quick overview of the 3 day short course.
Then we shall cover fundamental concepts of graph mining,
in 'recipe' format, and specifically:

(a) How to spot 'important' nodes in a network. We shall cover pageRank,
HITS, and we shall go deeper into singular value decomposition (SVD),
and the closely related eigenvalue analysis.

(b) How to find communities in a network. We shall mention 'METIS'
and 'cross-associations', and draw attention to the surprising
observation that social networks might *not* have good cuts.

(c) How to handle time-evolving graphs, using the powerful tool of tensors
(Parafac and Tucker decompositions)

For all the topics, we will emphasize the intuition,
and we will have exercises later in the afternoon.