Collection of Temporal Social Network Datasets and Replication Code for Delta Clustering Analyses

This archive contains a curated collection of five temporal social network datasets, together with replication code used to examine the robustness and generalizability of methods for analyzing dynamic local network structure. The datasets span several distinct social settings—including organizational email communication, smartphone-based interpersonal communication, face-to-face workplace contacts, and non-human social interaction—and vary substantially in network size, temporal resolution, and observational context. The collection includes the following datasets: Enron Email Dataset. This dataset represents email communication associated with the Enron Corporation. The original Enron email corpus was made publicly available following the Federal Energy Regulatory Commission investigation and subsequently prepared and distributed for research purposes. Email communication records can be represented as a temporal network in which nodes correspond to email addresses and interactions correspond to email communication events over time. The data provide a large-scale example of an organizational communication network. NetSense Dataset. The NetSense project is a longitudinal study conducted at the University of Notre Dame beginning with a cohort of incoming undergraduate students in Fall 2011. Participants were provided smartphones that recorded metadata on telephone calls and text-message exchanges. The communication data contain time-stamped records of interactions among study participants and can be used to construct longitudinal and temporal communication networks. NetSense also collected survey information on participants’ social relationships, activities, attitudes, and background characteristics; the temporal communication data are the component included in this collection. Contacts in a Workplace. This dataset contains a temporal network of face-to-face contacts measured among individuals working in an office building in France from June 24 to July 3, 2013. Contacts were detected using the SocioPatterns wearable-sensor infrastructure and recorded at 20-second temporal resolution. Anonymous participant identifiers can be linked to department information supplied with the original dataset. The data provide a high-resolution representation of the evolution of interpersonal contact patterns in a workplace. Contacts in a Workplace, 2nd Deployment. This dataset contains a second temporal workplace contact network collected in an office building in France in 2015. As in the earlier workplace deployment, face-to-face contacts were measured using the SocioPatterns infrastructure and represented as interactions between anonymous individuals during 20-second intervals. Departmental affiliation information is also available. The second deployment provides an independent workplace setting for examining the temporal organization of interpersonal contacts. Baboons’ Interactions. This dataset contains social-interaction data collected from a group of Guinea baboons (Papio papio) living in an enclosure at a primate research center in France. Data were collected between June 13 and July 10, 2019, using direct behavioral observations and wearable proximity sensors. The sensor data record time-stamped proximity contacts among equipped animals, providing a high-resolution temporal network of non-human social interactions. The dataset enables comparison of temporal network dynamics across human and non-human social systems. The archive also includes replication code for the analyses of temporal local clustering. The R code takes a temporal edge list in CSV format with three columns (i, j, and date) and constructs the local clustering and delta-clustering measures and their structural components, including persisting, new, and lost contacts and the corresponding subgraph densities. The Stata code estimates the fixed-effects models used to examine the association between these structural components and changes in local clustering. A README file documents the input format, analysis workflow, and use of the accompanying code. These datasets and replication materials are archived together to facilitate reproducibility, methodological reuse, and comparative analysis of temporal social networks. The original datasets were produced by different research projects and remain subject to their respective attribution requirements and licensing terms. Users of this archive should cite the original data sources and associated publications, as appropriate, in addition to citing this Zenodo record.