Description Details Author(s) References
Combine multiple-relationship networks into a single weighted network. The approach is similar to factor analysys in the that contribution from each constituent network varies so as to maximize the information gleaned from the multimetwork. This implementation uses Principal Component Analysis calculated using 'prcomp' with bootstrap subsampling.
Package: | dils |
Type: | Package |
Version: | 0.8 |
Date: | 2013-10-27 |
License: | MIT + file LICENSE |
Start with a table (data.frame, tab-delimited file, database) where each row/record represents a link between two nodes (a dyad) in a directed or undirected network and each column represents a different relationship between the two nodes, ie. each column is a network. DILS combines these columns/networks into a single network that is a weighted sum of the constituent networks. The resulting DILS network uses information from all of the constituent networks and contains more information than any of the constituent networks. The output is a data.frame of DILS scores for each dyad, therefore is a single network ready for analysis using igraph or other social network analysis (SNA) tools.
Workflow synthesizing networks might typically look like this:
Start with several networks in igraph, adjacency list, or edgelist form.
Is necessary, use EdgelistFromIgraph
or EdgelistFromAdjacency
to convert igraph and adjacency list networks to edgelist form.
Use MergeEdgelists
to combine the individual network datasets into a single dataset.
Use GenerateDilsNetwork
to synthesize the networks in the merged data set into a single weighted network.
Use IgraphFromEdgelist
or AdjacencyFromEdgelist
to convert the edgelist output to the desired output.
Use RelativeNetworkInformation
on input networks and DILS network to see if/how much the information content of the DILS network exceeds the information content of the input networks.
Workflow for imputing edges for a binary network might typically look like this:
Start with a binary network as an adjacency matrix (for an igraph use get.adjacency
).
Use RelationStrengthSimilarity
to Calculate RSS scores for each dyad.
Use RssSuggestedNetwork
on the original network and the RelationStrengthSimilarity output to get a new suggested network with more edges.
Stephen R. Haptonstahl <srh@haptonstahl.org>
"Discovering Missing Links in Networks Using Similarity Measures", Hung-Hsuan Chen, Liang Gou, Xiaolong (Luke) Zhang, C. Lee Giles. 2012.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.