Integrating Agent-Based Simulations with Archeological Fieldwork through Log-Data Analysis

Abstract

This chapter proposes a new framework that analyzes the log data from computational simulations to integrate agent-based simulation (ABS) with conventional fieldwork-based research in archeology. Specifically, machine learning methods are employed to extract the key branch points of each result from a large and diverse set of simulation results and the log data representing the process leading to them. In this framework, the ABS results are used to develop working hypotheses for facilitating fieldwork-based research, whereas those of fieldwork are used as inputs for the simulations, thereby resulting in bidirectional links rather than unidirectional ones. In a pilot application of this framework, the log data from a simulation of the cultural transformation from the Jomon period to the Yayoi period (16,000 to 2350 cal BP) in Western Japan are analyzed.