[gov3009-l] Blitzstein on "Stability and Estimability of Centrality in Networks"

Hello, We hope you will join us this Wednesday, March 24th at the Applied Statistics workshop when we will be happy to have Joe Blitzstein (Department of Statistics). Details and an abstract are below. A light lunch will be served. Thanks! "Stability and Estimability of Centrality in Networks" Joe Blitzstein Harvard Department of Statistics March 24th, 2010, 12 noon K354 CGIS Knafel (1737 Cambridge St) Abstract: "Centrality" is one of the most widely-used notions network analysis, attempting to measure the importance of a node. Yet despite being such a central idea to the study of networks, there is no standard definition; rather, there are many incompatible definitions, e.g., based on degrees, shortest paths, and eigenvectors of the adjacency matrix. We will attempt to sort through this menagerie of definitions of centrality from the perspective of stability and estimability. That is, if the network is perturbed slightly or is measured with noise, how much does centrality change? If -- as is usually the case -- only a sample of the full network is observed, which of these notions can be reliably estimated from the sample? Cheers, matt. ~~~~~~~~~~~ Matthew Blackwell PhD Candidate Institute for Quantitative Social Science Department of Government Harvard University email: mblackwell(a)iq.harvard.edu url: http://people.fas.harvard.edu/~blackwel/