***** To join INSNA, visit http://www.insna.org ***** I transformed the Panama Papers dataset into Pajek's multirelational network and I prepared an overall picture how different relations link different types of nodes. See http://vladowiki.fmf.uni-lj.si/doku.php?id=notes:net:pa From this network it is easy to extract different subnetworks. with best wishes, Vlado On 14-05-2016 15:55, Moses Boudourides wrote: > ***** To join INSNA, visit http://www.insna.org ***** > > Hello, > > I've been also spending some time with the Panama Papers dataset. > However, in what concerns the network structure that could be > extracted from this data set, it's not yet very clear to me which > relations could be used for that purpose. The relational part of the > dataset is the file all_edges.csv, which contains 3 columns: the first > and the third columns contain node_ids and the second refers to the > following five types of relations that associate a node of the first > column with the corresponding node of the third column: > 'intermediary_of', 'officer_of', 'registered_address', 'similar', > 'underlying'. Apparently only the fourth type ('similar') is symmetric > (undirected) and all the other four types are obviously directed. > > Given that the total number of all relations (edges) is very high > (1265690), I was wondering what sort of aggregations among types of > relations might simplify the complexity of the Panama papers network. > > I would appreciate if someone is willing to share any ideas about a > meaningful aggregation scheme for relations. Of course, one could > disregard any sort of relational aggregation and treat the network as > a multilayered (multiplex) one, although the size and the complexity > of the Panama Papers network appear to be rather restraining. > > I should add that, following Dmitry Zinoviev's original work on this > dataset, at the moment, I can make a number of computations and > visualizations for parts of the network (I'm using Python's Networkx > and Lightning-Python for interactive visualizations). For instance, > being motivated by what Dmitry is doing, I've managed to analyze the > ego-networks extracted from egos which are nodes of certain type > (officers, intermediaries, addresses, entities) associated with a > particular country and being connected to alters according to a > certain relationship type. > > For instance, this is the (symmetric) network in the case that egos > are Greek officers and alters correspond to international entities and > addresses (aggregated by all types of relations-edges): > > http://public.lightning-viz.org/visualizations/0a2669fc-f8bd-4cbf-b66f-1110f63c49df/public/ > > (This is just an example: I can produce such ego-centric networks for > any country in the Panama Papers data.) > > Admittedly, I'm not pleased with the aggregation of relations I'm > doing here (perhaps the inclusion of addresses was redundant too) and, > thus, I would ask for your ideas, comments or suggestions. > > --Moses -- Vladimir Batagelj IMFM - Institute of Mathematics, Physics and Mechanics Jadranska 19, 1000 Ljubljana, Slovenia and University of Primorska, Andrej Marušič Institute, Koper T: +386 1 4766 672 W: http://vladowiki.fmf.uni-lj.si/doku.php?id=vlado _____________________________________________________________________ SOCNET is a service of INSNA, the professional association for social network researchers (http://www.insna.org). To unsubscribe, send an email message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in the body of the message.