Virtual Kinship Networks: Exploring Social Networks from an Agent-Based Model

DSpace Repositorium (Manakin basiert)


Dateien:

Zitierfähiger Link (URI): http://hdl.handle.net/10900/146418
http://nbn-resolving.de/urn:nbn:de:bsz:21-dspace-1464182
http://dx.doi.org/10.15496/publikation-87759
Dokumentart: Konferenzpaper
Erscheinungsdatum: 2023-10-31
Originalveröffentlichung: Human History and Digital Future : Proceedings of the 46th Annual Conference on Computer Applications and Quantitative Methods in Archaeology
Sprache: Englisch
Fakultät: 5 Philosophische Fakultät
Fachbereich: Archäologie
DDC-Klassifikation: 930 - Alte Geschichte, Archäologie
Schlagworte: Soziales Netzwerk
Freie Schlagwörter: agentenbasierte Modellierung
exchange
social networks
agent-based modeling
Zur Langanzeige

Abstract:

This paper analyzes the structure of virtual kinship networks formed by an agent-based model that was originally designed to explore the relationships among kin networks, residence rules, settlement size, and the movement of exchange goods.Following simple rules, agents in the model are born, die, find mates, establish post-marital residence. Agents then exchange goods (which are conceptualized as pottery vessels) among close kin dispersed through a linear system of villages. Each run of the model produces a network that unites most agents, but each agent also has a personal network of close kin. Previous analysis of model output has focused on variation in the number of virtual pottery vessels obtained by agents, and on data averaged over large numbers of model runs, with only minimal analysis of the networks produced. But variation in network structure must underlie the variation in exchange success seen in the model runs. This paper focuses on the virtual networks produced by the model, including examining variation in measures of centrality and degree distribution, as well as variation in path length from one end of the system to the other. The data exploration reported here indicates that centrality is important, but centrality alone is not a good predictor of success in exchange. Agents who obtain large numbers of vessels typically are connected to producers directly or through one or two intermediate links, and also tend to have relatively high centrality in the network.

Das Dokument erscheint in: