Element-level Analysis (google's Pagerank)

Many readers will find the title of this book misleading – at least, at first sight. This is because ‘network’ is a heavily overloaded term used to denote relational data in so vast a number of applications that it is far from surprising that ‘network analysis’ means different things to different people. To name but a few examples, ‘network analysis’ is carried out in areas such as project planning, complex systems, electrical circuits, social networks, transportation systems, communication networks, epidemiology, bioinformatics, hypertext systems, text analysis, bibliometrics, organization theory, genealogical research and event analysis. Most of these application areas, however, rely on a formal basis that is fairly coherent. While many approaches have been developed in isolation, quite a few have been re-invented several times or proven useful in other contexts as well. It therefore seems adequate to treat network analysis as a field of its own. From a computer science point of view, it might well be subsumed under ‘applied graph theory,’ since structural and algorithmic aspects of abstract graphs are the prevalent methodological determinants in many applications, no matter which type of networks are being modeled. There is an especially long tradition of network analysis in the social sciences [228], but a dramatically increased visibility of the field is owed to recent interest of physicists, who discovered the usefulness of methods developed in statistical mechanics for the analysis of large-scale networks [15]. However, there seem to be some fundamental differences in how to approach the topic. For computer scientists and mathematicians a statement like, e.g., the following is somewhat problematic.