Marco Valenti, Girolamo Fiorentino, Stefano Bertoldi, Miriana Concetta Colella, Anna Maria Grasso, Cristina Menghini, Alessandra Nardini, Carla Palmas, Manuele Putti, Dal caos all’ordine? Sperimentazioni di Social Network Analysis e Intelligenza Artificiale sulla cultura materiale del villaggio di Miranduolo (secoli VIII-X)

The site of Miranduolo (Chiusdino – Siena) was excavated by the University of Siena between 2001 and 2017, covering an area of approximately 6,000 square meters—nearly the entire stratigraphic deposit within the stone castle walls. These investigations have made it possible to understand the phases of occupation on the hilltop from the 7th to the early 14th century. This paper aims to study the site using innovative approaches, analyzing the complex dataset of artefacts and ecofacts dating from the 8th to the 9th–10th centuries, and to interpret the settlement transformations detectable in those phases. During the 8th century, Miranduolo appears as a village structured around two distinct possessores, along with two additional areas occupied respectively by a blacksmith and a figure resembling a land manager and a cluster of huts belonging to dependent peasants. Between the 8th and 9th centuries, the topographic and socio-political configuration of the site changes, as one of the two possessores consolidates control—possibly through the intermediation of an actor—over the entire population of farmers and craftsmen. The archaeological research has produced a vast amount of information, published in numerous academic venues. This includes a heterogeneous set of pottery, metal, glass, archaeobotanical, and archaeozoological finds totaling over 84,000 records, quantified by Minimum Number of Individuals (MNI) rather than by fragment counts—an amount, while not reaching Big Data scale, that is nonetheless difficult to process without advanced digital and statistical methods. The aim is to reconstruct the social structures of early medieval Miranduolo and the economic and political relationships among its inhabitants through tools such as Social Network Analysis (SNA) and Artificial Intelligence (AI), comparing the results with those obtained through traditional stratigraphic data analysis. Ultimately, the study seeks to understand how power shaped external supply systems, agricultural production and animal husbandry, surplus storage, and even the topography of the hill and the composition of its population. SNA will be used to assess levels and patterns of material-based interconnection among the various areas of the 8th-century poly-nuclear settlement and the 9th–10th-century manorial village, while the computational and analytical power of AI large language models (LLMs) will help quantify the social roles and wealth of the inhabitants. This experimental approach offers an alternative method for evaluating the validity of previous interpretations and archaeological models, potentially correcting certain aspects or proposing new perspectives for understanding and reconstructing the early medieval society of Miranduolo.