The Architecture of Nonverbal Communication: Data and Code for a Bipartite Network Analysis of Cue-State Relationships in Learning

Zenodo record description (anonymized for review) Data and analysis code for a bipartite network analysis of cue-state relationships in learning. The network comprises 2,010 internal-state nodes, 6,434 behavioral-cue nodes, and 10,240 weighted edges derived from a systematic review of 908 empirical studies (1966-2025). The pipeline constructs the bipartite network and its unipartite projections; computes global metrics and heavy-tailed degree-distribution fits; identifies hub cues and states via multiple centrality measures; detects community structure with the Louvain algorithm; quantifies the generalist-specialist spectrum of cues through a Generalist-Specialist Index; and relates evidence strength (replication tiers R1-R6) to network position, including a disparity-filter backbone. All results are exactly reproducible under a fixed seed. The deposit includes the full Python pipeline, the input relationship data, and the generated figures. This record accompanies a manuscript currently under double-anonymous review. All author names, affiliations, contact information, and identifying citations have been withheld to preserve review anonymity; this metadata will be added upon acceptance. Released under the Creative Commons Attribution 4.0 International License (CC BY 4.0).