abstract |
Are two given networks similar? Are there similar parts in a network? How robust is a network to failures? - Mining and analyzing very large graphs is becoming increasingly important, e.g. in the field of Internet search engines. However, most network analysis and mining methods do not scale up well to instances with millions of nodes. This talk presents the ANF algorithm proposed by Palmer et al., an analysis tool designed to be fast, accurate, and scalable to handle large disk-resident data sets. It computes an approximation to distance degree sequences, which can be seen as "reachability profiles" for both the particular nodes and the whole graph. The authors claim that the results of their algorithm can help to efficiently answer the abovemetioned questions, some samples of which are discussed in the talk.
|