The Show Programme, the LLM and the Ontology : Designing Workflows for Performing Arts Analytics Communication dans un congrès

Theatre programmes are a vital resource for reconstructing the conditions of a performing arts production. Found across numerous heritage collections, they provide direct data on performers, technicians, and venues while revealing artistic networks, editorial strategies, and institutional dynamics. Although digitisation has enhanced access to these often-overlooked materials, significant methodological challenges persist : inconsistent database transcriptions, heterogeneous data models with limited interoperability, and no reliable link to the original source. Drawing on the ERC project From Stage to Data, which examines programmes from the Festival d’Avignon (1947–present), this talk addresses key methodological challenges in performing arts analytics. This presentation outlines a computational workflow that combines Large Multimodal Models (LMMs) with an extension of the Linked Art ontology for performing arts. The extracted data are connected to its source via the Arvest platform, ensuring full traceability. Examples from the Avignon corpus demonstrate how this approach enables analyses of artistic circulation and collaboration over time. This presentation also calls for the creation of a comprehensive corpus of theatre programmes, which would unlock numerous research avenues in performing arts history. Such a corpus would provide the research community with standardized, interoperable data, supporting comparative studies across institutions, time periods, and geographical contexts.