Abstract
This study aims to investigate the dynamics of obsidian trade networks during the Jomon period (15,000–2,400 years cal BP), the hunting and gathering era in Japan. To improve regional representation and reduce the distortions caused by small sample sizes, we performed clustering based on a large-scale dataset and conducted social network analysis. The research results revealed that the trade networks during the Jomon period were not constant; they expanded throughout the southern Kanto region during the Middle Jomon period (5,500–4,500 years cal BP) and ceased to function during the Late Jomon period (4,500–3,200 years cal BP). Furthermore, to enhance the readability and interpretability of the dataset, we implemented clustering using the density-based spatial clustering of applications with noise (DBSCAN) method. The results showed that in every time division of the Jomon period, the mean intra-cluster cosine similarity of each cluster was higher than the similarity between sites outside the clusters, confirming the reasonableness of an analysis considering regional representation. In addition, to verify the robustness of the network in the social network analysis after clustering, we also performed a bootstrap simulation analysis. The results showed high network robustness and demonstrated that the sampling after clustering had minimal impact on this study’s findings.