The “Self Organization” network represents a network extracted from 572 entries of the ISI Web of Knowledge
. Such corpus corresponds to the articles published in France between 2006 and 2010 and containing the keyword "self organization".
From the metadata of each of these entries, 3 types of items have been extracted corresponding to the 3 types of nodes in the graph: the keywords used by each article (blue), the references quoted by each article (red) and the institutions hosting the authors of each article (green). For the sake of legibility, the graph contains only the most occurring elements in our corpus: i.e. 373 keywords, 572 references, 187 institutions.
Edges represent co-occurrence in the same entry. E.g. two articles are connected if they are both quoted by a third article. Two keywords are connected if they are both associated with the same article. Two institutions are connected if they host the authors of the same article. A keyword is connected to an institution if they are both mentioned in the same article. An article is connected to a keyword, if the article is quoted by another article associated with the keyword. An institution is connected to a keyword if the are both mentioned in the same article
The network has been spatialized using the software Gephi (gephi.org) and the force vector algorithm ForceAtlas2.
The image represents a network extracted from the 1893 entries of the ISI Web of Science, corresponding to the articles published between 2008 and 2010 and containing the keyword “self organization”. From the metadata of each of these entries, 3 types of items have been extracted corresponding to the 3 types of nodes in the graph: the keywords used by each entry (green), the references quoted by each entry (red) and the institutions hosting the authors of each entry (blue). For the sake of legibility, the graph contains only the most frequent nodes (keywords and references appearing more than 12 times and institutions more than 6 times), leading to 125 keywords, 172 references, 206 institutions.
Lines represent co-occurrence in the same WoS entry and connect nodes appearing in the same WoS entry, no matter their type. The network is spatialized using the software Gephi (gephi.org) and the force vector algorithm ForceAtlas2 (17).
The colors in the image are computed as as a density heatmap. Lighter areas of the image correspond to regions of the graph with a higher density of nodes. The image is constructed by overlapping three density heat maps of three different colors: green for keywords, red for references, blue for institutions. The color of each point of the image is calculated according to the light radiating from all nodes (each node radiates according to an exponential decay whose half-life is proportional to its in-degree).
Colors have been aggregated in nine thresholds of equidistant luminosity, to improve the legibility of the image. Since the colors employed are pure (0000FF for blue, FF0000 for red; 00FF00 for green) their luminosity adds up mathematically in the RGB image: e.g. light violet areas are dense in institutions and references.
(Designed by Sciences Po médialab, Benedetta Signaroldi & Daniele Guido)