Jeudi 23 avril 2015 à 11h, salle 25-26/101
As we recognise the importance of information in graph or network format it is useful to be able to quantify how similar one graph is to another. This is particularly useful in the context of ego networks, the local networks around a nodes. This seminar focuses on network motif profiles (i.e. counts of specific network motifs) to characterise networks in much the same way that a bag-of-words strategy allows text documents to be compared in a vector space framework. This strategy has a long history in social science and bioinformatics and has considerable potential in social network analysis. The seminar will outline the general strategy and present case-studies from Wikipedia, YouTube spam campaigns and peer to peer lending in the Prosper marketplace.