Coupling Analysis
It performs a coupling network analysis and plots community detection results on a bi-dimensional map (Coupling Map).
couplingMap( M, analysis = "documents", field = "CR", n = 500, label.term = NULL, ngrams = 1, impact.measure = "local", minfreq = 5, community.repulsion = 0.1, stemming = FALSE, size = 0.5, n.labels = 1, repel = TRUE, cluster = "walktrap" )
M: is a bibliographic dataframe.analysis: is the textual attribute used to select the unit of analysis. It can be analysis = c("documents", "authors", "sources").field: is the textual attribute used to measure the coupling strength. It can be field = c("CR", "ID","DE", "TI", "AB").n: is an integer. It indicates the number of units to include in the analysis.label.term: is a character. It indicates which content metadata have to use for cluster labeling. It can be label.term = c("ID","DE","TI","AB"). If label.term = NULL cluster items will be use for labeling.ngrams: is an integer between 1 and 4. It indicates the type of n-gram to extract from texts. An n-gram is a contiguous sequence of n terms. The function can extract n-grams composed by 1, 2, 3 or 4 terms. Default value is ngrams=1.impact.measure: is a character. It indicates the impact measure used to rank cluster elements (documents, authors or sources). It can be impact.measure = c("local", "global").\ With impact.measure = "local", couplingMap calculates elements impact using the Normalized Local Citation Score while using impact.measure = "global", the function uses the Normalized Global Citation Score to measure elements impact.minfreq: is a integer. It indicates the minimum frequency (per thousand) of a cluster. It is a number in the range (0,1000).community.repulsion: is a real. It indicates the repulsion force among network communities. It is a real number between 0 and 1. Default is community.repulsion = 0.1.stemming: is logical. If it is TRUE the word (from titles or abstracts) will be stemmed (using the Porter's algorithm).size: is numerical. It indicates the size of the cluster circles and is a number in the range (0.01,1).n.labels: is integer. It indicates how many labels associate to each cluster. Default is n.labels = 1.repel: is logical. If it is TRUE ggplot uses geom_label_repel instead of geom_label.cluster: is a character. It indicates the type of cluster to perform among ("optimal", "louvain","leiden", "infomap","edge_betweenness","walktrap", "spinglass", "leading_eigen", "fast_greedy").a list containing:
map | The coupling map as ggplot2 object | |
clusters | Centrality and Density values for each cluster. | |
data | A list of units following in each cluster | |
nclust | The number of clusters | |
NCS | The Normalized Citation Score dataframe | |
net | A list containing the network output (as provided from the networkPlot function) |
The analysis can be performed on three different units: documents, authors or sources and the coupling strength can be measured using the classical approach (coupled by references) or a novel approach based on unit contents (keywords or terms from titles and abstracts)
The x-axis measures the cluster centrality (by Callon's Centrality index) while the y-axis measures the cluster impact by Mean Normalized Local Citation Score (MNLCS). The Normalized Local Citation Score (NLCS) of a document is calculated by dividing the actual count of local citing items by the expected citation rate for documents with the same year of publication.
## Not run: data(management, package = "bibliometrixData") res <- couplingMap(management, analysis = "authors", field = "CR", n = 250, impact.measure="local", minfreq = 3, size = 0.5, repel = TRUE) plot(res$map) ## End(Not run)
biblioNetwork function to compute a bibliographic network.
cocMatrix to compute a bibliographic bipartite network.
networkPlot to plot a bibliographic network.
Useful links