Browse Books

Reviewer: Yousri ElFattah

The authors claimed in the first edition of this book that it was "the first textbook on social network analysis [SNA] integrating theory, applications, and professional software for performing network analysis" [1]. Pajek is a computer program for SNA that plays a central role in the book's examples and illustrations. The content of this second edition differs from the first edition in that it has additional Pajek updates, expanded references and listings for further reading, and a new chapter 13 on random graph models. This book is divided into five parts. Part 1 is a foundational introduction to exploratory SNA using Pajek. Parts 2 to 5 focus on specific network topics. Part 2, "Cohesion," has chapters on cohesive subgroups, sentiments and friendship, and affiliation. Part 3, "Brokerage," includes chapters on center and periphery, brokers and bridges, and diffusion. Part 4, "Ranking," has chapters on prestige, ranking, genealogies, and citations. Part 5, "Roles," has chapters on block models and random graph models. Each chapter ends with assignments and exercises with their answers, as well as recommendations for further reading with bibliographical notes. The chapters are interspersed with application sections, which include examples that can be run using Pajek; such hands-on experience reinforces learning. SNA is a formal methodical analysis that deals with the exploration and visualization of social network structure and data, as well as with the testing of hypotheses by means of statistical models. This book is focused on exploratory analysis, as reflected in its title, and considers statistical analysis outside of its scope. The input data for exploratory SNA is expressed formally with annotated graphs (directed or undirected) consisting of nodes (representing individual actors within the network) and ties (representing social relationships between the actors, which are social entities such as people, organizations, business units, and countries). Examples of social relations include kinship, affection, interaction, and affiliation. Pajek is a network drawing tool with network analysis capabilities that include calculating centrality measures, identifying structural holes, and visualizing block models. Centrality measures indicate the concentration of ties on a limited number of actors in the network. Structural holes indicate the extent to which an actor connects other actors that are not connected directly. Block models group or block entities into types and analyze the network structure with respect to types rather than individual actors. I view Pajek more as an experimental tool for academic use and less as a polished tool suitable for commercial or business use. It lacks an application programming interface (API) or a scripting language to customize performance. Pajek is a menu-driven graphical user interface (GUI) requiring many manual steps to transition from input data to analysis results. The menus offer the user basic graph operations on one or two networks that need to be stitched together for custom analysis, which requires a steep learning curve. Macros can be recorded to perform repetitive tasks, but this is not much help. Pajek has a feature to send data directly to the statistical analysis tools R or SPSS, but the linkage with statistical models is not sufficiently elaborated on in the book. Pajek has graph drawing capabilities for network visualization, but users can view only one network at a time. This makes it difficult to visualize the effect of network manipulations, as one cannot see the networks side by side, before and after being operated upon. Also, the content of the input file cannot be viewed or edited from within Pajek. This is a significant limitation because the encoding network has many syntax varieties, adding cognitive burdens on the user to keep track of the types of networks loaded and the queries supported by their structures and data models. The treatment of the topics in the book is very informal and more oriented to social science readers than computing professionals. There is little mention of data structures and algorithmic aspects of SNA, and there are hardly any computational complexity statements and empirical evaluations of various social analysis tasks on varying random graphs. It would be useful to identify what structural parameters in a social network impact the various analysis tasks. The authors boast that "Pajek can handle networks up to 999,999,997 vertices" (Appendix A, p. 381). They should substantiate or qualify such a statement and explain what "handle" means in this context. This book is clearly intended as a textbook for a classroom setting, but its relevance to a wider audience could be enhanced if the application sections included motivating real-life use cases with sample real-life queries. I view the book more as a tutorial for a graphical and analytical tool (Pajek) than as a general resource for gaining experience and insight in the important field of SNA. Overall, the book is an important resource on exploratory social analysis techniques in general, and I recommend it as a textbook for a course on the topic. Online Computing Reviews Service