networkflow: a research pipeline for projected and dynamic networks

networkflow organises the study of networks built from tabular research data (articles and their references, authors and their papers, accounts and what they share) as a sequence of explicit, documented steps. Each step is a function whose choices are arguments that can be changed and reported, and each returns a tbl_graph (or a list of them, one per time window) that the next step enriches.

Details

• Building networks: build_network(), build_dynamic_networks() (bibliographic coupling, co-citation, coupling of authors and other projections of two-mode tables, with similarity measures from biblionetwork or statistical backbones from backbone, one network per time window), filter_components().

• Finding and reading clusters: add_clusters() (Leiden, Louvain, Infomap...), merge_dynamic_clusters() (following clusters across windows), merge_diagnostics() and plot_merge_diagnostics() (how the result depends on the threshold), name_clusters(), extract_tfidf(), add_node_roles().

• Drawing and exploring: layout_networks(), color_networks(), prepare_label_networks(), plot_networks(), networks_to_alluv(), minimize_crossing_alluvial(), plot_alluvial().

The guides of the package website (https://agoutsmedt.github.io/networkflow/) explain the methods behind each step on the example of the literature on the 1970s stagflation (Nodes_stagflation, Ref_stagflation).

See also

Useful links:

• https://github.com/agoutsmedt/networkflow

• https://agoutsmedt.github.io/networkflow/

• Report bugs at https://github.com/agoutsmedt/networkflow/issues

Author

Maintainer: Aurélien Goutsmedt agoutsmedt@hotmail.fr (ORCID)

Authors:

• Aurélien Goutsmedt agoutsmedt@hotmail.fr (ORCID)

• Thomas Delcey (ORCID) (co-maintainer)

• Alexandre Truc (ORCID) (initiated the alluvial method)