Creating a Dynamic Knowledge Graph of the Arabic Manuscript Tradition
Conference presentation by Daniel Kinitz at the Deutscher
Orientalistentag 2022
How can we generate and integrate information on manuscripts in such a way that research can gain new insights? Bibliotheca Arabica’s digital research platform aims to support researchers to shed new light on the production, transmission, and reception of Arabic literatures from the 12th to the 19th century. The heterogeneity of the sources - especially manuscript catalogues, reference works and manuscript notes - creates the need for a flexible approach to generate and process manuscript-related data.That is to say, a research platform has to be more than a static repository, granting access to more or less known facts. Rather, creating a knowledge graph for the Arabic manuscript tradition requires a dynamic, multi-layered modelling of data as well as a flexible workflow of data integration.
This includes the retroactive, increasing tapping of sources, implemented through a flexible data extraction process, a permanent enrichment of existing data, e.g., cross-linking entities, as well as flexible authority control, taking new findings into account. No fixed input mask is used, but rather different layers of data (catalogue layer, processing layer, output layer) are being integrated, thereby allowing for contradicting fact(oid)s and differing categories. While recording and cross-linking entities like works, manuscripts, and related agents, the platform makes the provenance chain of each factual statement transparent and enables researchers to assess the quality of a specific source.
The talk will present approaches and best practices in the making and show relevant features of Bibliotheca Arabica’s new research platform.