The Dataverse Network Project

You can analyze network data sets by using the Network Measure options on the Data File page. These analysis tools collect measures of centrality, or importance, of the main actor in a network data set. Network measures are calculated for every element and appear as new attributes in the graph, and can be used in queries like any other attribute. See Network Data Tips for specific details about how to perform network measure analysis.

Note: Only network measures that are vertex attributes are supported.

The Dataverse Network uses the igraph library to perform the network measures analysis. The following measures are available:

• Page rank - This is the PageRank measure as defined by Page and Brin, the founders of Google, in 1998.¹
  PageRank is a measure of centrality in which a particular vertex's centrality depends on the centrality of its neighbors. PageRank uses the damping parameter <d>, which defines what proportion of a vertex's centrality is derived from its neighbors and what proportion arises exogenously.
• Degree - This is the number of edges that touch a vertex. This is a first-order way to see how connected a particular vertex is in the graph.
• Unique degree - This is the number of unique vertices to which a particular vertex is connected directly. In a network where there can be only one edge between two vertices, degree and unique degree are identical. In a network where multiple edges are allowed between two vertices, unique degree is less than or equal to degree.
• In largest component - This is a simple flag indicating whether a vertex is in the largest connected component of the graph.
• Bonacich Centrality - This is the alpha centrality measure defined by Bonacich and Paulette in 2001.²
  Alpha centrality is another measure in which a particular vertex's centrality depends on the centrality of its neighbors. Alpha centrality makes use of two parameters. Alpha defines how much a neighbor's centrality rubs off on or detracts from a particular vertex's centrality. Exo defines how much centrality arises exogenously from the structure of the graph.
  Only uniform exogenous centrality is supported.

¹Brin, S. and Page, L. (1998). The Anatomy of a Large-Scale Hypertextual Web Search Engine. In: Seventh International World-Wide Web Conference (WWW 1998), April 14-18, 1998, Brisbane, Australia. http://infolab.stanford.edu/~backrub/google.html.

²Bonacich, P. and P. Lloyd (2001, July). Eigenvector-like measures of centrality for asymmetric relations. Social Networks 23 (3), 191-201.