Pandora & IsoMemo Initiatives for Big Historical Data Compilation and Modelling
An effective understanding of past historical dynamics under a systems approach requires large volumes of diverse data. This should be structured as linked open data so that different systems’ components can be efficiently connected. To achieve this, we developed the Pandora data platform in collaboration with the Max Planck Computing and Data Facility and the Max Planck Library plus c. 60 international partners. Pandora is a grassroots initiative promoting the creation of independently managed data communities and wider collaborative data networks. Among these is the IsoMemo network connecting c. 40 open-access databases that compile isotopic data informing on the environmental history of the Earth and of its human inhabitants from the deep past until the present day. Amassed data via Pandora is used to test historical hypotheses and in data-driven approaches that seek to identify historical patterns.
We are also developing novel R modelling tools for historical research. These are made available online via the Pandora & IsoMemo open-source software platform and are designed to address typical aspects of historical data such as missing data, data uncertainties, and reliance on expert information. We employ Bayesian modelling of proxy data and expert inputs to produce high-resolution reconstructions of various aspects that characterized past human-environmental systems. Machine learning techniques, such as Bayesian networks and non-linear model selection algorithms, are used to study historical causation. We develop and/or collaborate in the development of models and software to investigate various aspects of complex systems such as the study of cyclical human and environmental phenomena, detection of critical transitions and their warning signs, simulation of emerging phenomena using agent-based modelling, and in network analysis.
The Pandora & IsoMemo initiatives are also involved in direct data collection efforts although this is mostly carried out by network members. There are several research projects underway that apply newly developed modelling tools to compiled datasets. This includes projects investigating the spread of peoples, domesticates, and technologies at continental scales; determining the causes of species extinction; and surveying un/successful human adaptions to climate change.
