SOCIAL NETWORK ANALYSIS FOR DUMMIES YANNE BROUX DH
SOCIAL NETWORK ANALYSIS FOR DUMMIES YANNE BROUX DH SUMMER SCHOOL LEUVEN, SEPTEMBER 8 2015
TERMINOLOGY
Useful sources A. -L. BARABÁSI, Linked: The Science of Networks (Cambridge, 2002) S. BORGATTI et al. , Analyzing Social Networks (L. A. , 2013) Y. BROUX & S. VANBESELAERE, Six Degrees of Spaghetti Monsters (spaghetti-os. blogspot. com)
Basics • Node (vertex) • Edge (tie) – Undirected – Directed – Weighted (valued) B A E C F D • Degree: how many edges to a node – Undirected: count edges – Directed: indegree vs outdegree
DATA MANAGEMENT
Adjacency matrix • Symmetric, binary e. g. who knows who • Symmetric, weighted e. g. distance between places
Adjacency matrix • Asymmetric, binary e. g. choose 3 friends to sit with • Asymmetric, weighted e. g. number of emails sent to colleagues
One-mode vs two-mode • 1 -mode: direct ties between actors (= adjacency matrix) • 2 -mode: ties between different entities (= affiliation matrix)
Adjacency vs attribute matrix • Adjacency matrix: only records ties between nodes • Attribute matrix: each column is different attribute of the nodes (gender, role, ethnicity, status, …) = ‘nodelist’ (vs ‘edgelist’)
Attribute matrix (nodelist)
SOFTWARE
UCINET + Netdraw + Almost anything you need for SNA is in here, very advanced statistics + All statistics can be loaded into visualization package + Free (re-download after each trial version) - Can’t handle large datasets (3000+ nodes) - Only for Windows (Mac: run with Wine but not all features work) - Crude visualizations
R + Very, very comprehensive + Not only SNA, everything statistical + Free + Open source - Steep learning curve (programming language) - Difficulty with Big Data - Very crude visualizations
Node. XL + Fancy visualizations + Easy interface (integrated in Excel!) + Free - Doesn’t work for Mac
Gephi + Free + More sophisticated visualizations + Easy import from Excel - Less (accurate) possibilities for analysis - Some bugs when reopening saved files - No user guides. You’re on your own not anymore! spaghetti-os. blogspot. com
SNA AND TRISMEGISTOS
Late Republican affairs
Co-publication
Names – Hermopolis AD 101 -130 322 names (based on 621 individuals) 471 edges Greek Egyptian Latin unknown
Place names in Egypt
year text person Threemode network
GETTING STARTED WITH GEPHI
Nodelist
Edgelist
Communities
‘Modularity’: automated community detection Density within clusters vs between them
Centrality measures A = betweenness B = closeness C = eigenvector D = degree
Centrality measures
Page. Rank
- Slides: 29