In our increasingly complex world, we are witnessing an exponential growth in the treasure trove of human experience. To access and utilise the entirety of data and information on the collective heritage of humanity, traditional disciplinary knowledge needs to be directed toward a formal science of heritage.
This transdisciplinary science will focus on how historical information on intercontinental trade, diplomacy, conflicts and other interactions among cities, nations and continents—now encoded in complex interactions of written, pictorial, sculptural, architectural and digital records, as well as oral memories, practices and performed rituals—can be organised and processed with the assistance of machine learning algorithms.
Unlocking historical data with artificial intelligence
Engineering historical memory is a methodology for the organisation of historical information in the machine learning age that I am developing in collaboration with Assoc Prof Cheong Siew Ann of NTU’s School of Physical and Mathematical Sciences, and Assoc Prof Erik Cambria and Asst Prof Joty Shafiq Rayhan of NTU’s School of Computer Science and Engineering.
Initially based on studies of chronicles and diaries from Venice, Italy (covering the years 1205-1433), and world maps (e.g. the Fra Mauro map, generated around 1450 in Venice), we recently included other coeval historiographical traditions, including Chinese, Greek, Russian, Malay and Arab sources, in the project.
In particular, we are applying computational techniques such as pattern recognition, data mining, machine learning algorithms derived from other disciplines, knowledge aggregators, as well as interactive and visualisation solutions.
The research project is now linked to the Laboratory for Interdisciplinary Bookish and Experimental Research (LIBER), which has been recently established in the library of the School of Art, Design and Media. The aim is to provide data-driven agent-based modelling and simulations for the study of the Afro-Eurasian communication networks.
Learning from the great world explorers
Our project aims to set the stage for an international laboratory that can address a significant gap in integrating and sharing historical data and knowledge.
By extracting information from primary narrative sources, such as Venetian and Malay chronicles, world maps—for instance, the 1457 Genoese world map and Fra Mauro’s mappa mundi—and travel accounts from explorers like Marco Polo, Ibn Battuta and Zhang He, we are able to generate bottom-up mathematical models for trade, conflict and diplomacy.
The data is then mapped to an electronic database and used in analytical environments to build linkages between parsed texts and recognised entities from other heterogeneous sources (e.g. Wikipedia and Open Street Map) and search engines (e.g. Google Scholar and Microsoft Academic).
The research team is now developing a web-based platform that allows “reading” of the 1457 Genoese world map and the Fra Mauro map—originally in Latin and Vernacular Venetian—in English, Mandarin and several other languages (Figure 1). This visualisation is empowered through real-time connection to Wikipedia, providing access to further primary and secondary literature sources.