SNA | Merging and Emerging Order

Topic: [Pajek] merging two datasets.

Above: Maximum flow = 40 within a single network connecting Jornal GGN and the Instituto Millenium,

Wouter de Nooy, co-author of the authoritative textbook on the social network analysis software package Pajek, answers a question I have had in mind myself.

Elisa, Because the two networks that you want to merge have different sets of vertices, you cannot use the Union of Lines and Intersection commands. Use the Union of Vertices to create a network with all vertices and lines from both networks. Now two vertices may refer to the same person or organization. Then use the Net>Partition>Vertex [Labels] command to obtain a partition with a class for each set of vertices that have exactly the same label. Finally, shrink vertices within these classes to one new vertex with the Operations>Shrink Network>Partition command.I think this will do the job but the resulting network may not be recognized as a two-mode network by Pajek any longer.

Let us suppose that the sum of the sum of vertices in the two networks is 69, or 36+33 but that a certain number of these labels refer to identical nodes. Pajek computes the intersection of these partitions and creates another just to hold the vertex names.

Redundant vertices are marked with hash-tag, #, otherwise used to indicate a node is representative of a cluster or reduction.

Note that this can be applied to cases of multiple overlapping edges and can be iterated in order to discover information about a relationship from various sources at various times.

Here, I am using a network crawled by WIRE and seeded with links from Arts & Letters Daily. I have another based on the membership rolls of the World Association of Newspapers, and finally a larger (400,000 URLs) network derived from a much more thorough crawling based on the results of the previous two investigations.

An effective way to develop seeds for your crawl is to search for “newsstands” and “rosters” — systematic lists representing explicit ties — or chances to create them efficiently.

I am also beginning to be able to recognize examples of network structure, such as brokerage roles.

 

From a reading of the CENPEC program Web site and an analysis of the roster of Abong, the Association of Brazilian NGOs, I could see that this diagram — created in yEd — was was an important one: a cut-vertex without which the analyst loses track of the institutional context.

In the network of Brazilian labor federations, above, I found IREX in an interesting position — a classic bridging structure, in this case between the U.S. diplomatic establishment and the policy research arm of the dominant labor federation, CUT.

George Soros and Bill Gates certainly do seem to be involved in a lot of “public diplomacy” offensives.

This is interesting because among other “roster” projects, I have been working on Brazilian labor and industrial federations. (The National Industrial Confederation — CNI — has ruined things by migrating to a new URL that has yet to percolate up.)

 

 

 

 

 

I have also been accompanying the emergence of strategic alliances among South American dailies — Mexico’s Jornada, for example.

 

 

The so-called Brazilian “dirty blogs” are exhibiting closer coordination with the alternative press and coming into their own as a genuine alternative. The mainstreaming of Brasil 247 is an example, as is the commitment of the iG group to venture cap and innovation.

I am sorry not to have prepared a more thorough presentation.