Developing a Data Model for Theatre Productions in a Wikibase Instance: A Case Study Approach

(1) Context and motivation

The transitory nature of a theatre production, in that it consists of fleeting events through its performances, cannot be denied. Yet it should be possible to preserve more of a production than its material remains in the form of video recordings, director’s books, photos, or theatre reviews.

This research project emerged from the observation that various initiatives in digital humanities and theatre studies are increasingly using digital methods to process theatre data in diverse ways and for different purposes. However, a clear gap remains in this field of research: existing approaches primarily focus on personnel, institutions, or textual sources of a production, while detailed data on staging, scenography, and set design elements, for example costumes, and stage components and their arrangements, has rarely been modelled.

In response, this research project is dedicated to these production-related data and characteristics. It investigates to what extent production characteristics can be modelled to make productions analysable and comparable, both retrospectively and from multiple perspectives. A corresponding data model was developed and implemented in a Wikibase instance using three selected productions as case studies to iteratively refine and validate the conceptual structure. In the long term, this Wikibase instance is planned to be enriched with further productions, so that the analytical potential of the system will increase as the dataset grows.

The unique contribution of this research approach lies in filling a methodological and technical gap in theatre studies by enabling the structured representation of scenographic and production elements in a semantic web-compatible environment. By making production characteristics accessible and queryable, the project enhances theatre scholarship and neighbouring disciplines, offering analytical opportunities not supported by previous research practices. Traditionally, knowledge about a production can be accessed only through ephemeral viewing experiences, personal notes, or limited archival materials; this approach enables students and researchers to access and analyse concrete staging details long after the performance has ended, in addition to attending a live theatre visit, facilitating more transparent, verifiable, and richly contextualised interpretations.

(2) Current research landscape at the intersection of Digital Humanities and Theatre Studies

As Estermann (2020, p. 33) states in his contribution on a Linked Open Data Ecosystem for the Performing Arts: “When it comes to the actual data, linked data publication in the area of the performing arts is still in a pilot phase”. A year later, Bardiot (2021, p. 100) similarly observes: “We are only at the beginning of the shift from theatre studies to ‘digital theatre studies’ – what we call theatre analytics. From the 1960s to the present day, theatre studies have been in the minority in the field of digital humanities”.

While the field has evolved rapidly, leading to numerous conferences and digital projects (Nikitas, 2023), theatre studies are not (yet) as represented within the Digital Humanities as other disciplines.

One such project is, for example, (Re-)Collecting Theatre History.¹ It developed methods to catalogue theatre-historical materials in accordance with international standards, integrating authority data and controlled vocabularies (Probst & Pinto, 2020, p. 163). Its database links objects, agents, and productions, allowing playbills to be linked to corresponding performances and participating artists.

Likewise, the Performing Arts Specialised Information Service,² brings together museums, library and archive collections with focus on the performing arts in a research and reference portal (Voß, 2017). The project portal provides metadata on objects such as playbills, programmes, posters, photographs, and videos, information on more than 130,000 individuals and 180,000 events (Fachinformationsdienst Darstellende Kunst, n.d.).

Beyond Germany, databases such as Theadok,³ and IbsenStage⁴ collect and publish metadata on specific performances and productions (Maignant et al., 2022; Theadok.at, n.d.). In the Swiss context, the SAPA Foundation preserves Switzerland’s cultural heritage of the performing arts, provides access to archival holdings of national significance, and promotes research through its Swiss Performing Arts Platform (SAPA Foundation, n.d.).

The WikiProject Performing Arts,⁵ coordinated by the Canadian Arts Presenting Association (CAPACOA) since 2020, seeks to create the world’s most comprehensive and highest-quality open database for the performing arts. It covers theatre, dance, cabaret, and musical theatre (Wikidata, 2023). Within the framework of this project, several Wikidata items have already been created, for example, relating to individuals and their data, or to theatre forms, theatre artifacts, and organizations (Wikidata, 2022, 2025).

In addition to projects working with production metadata, there is a project that digitised, transcribed, and in 2016 online published the director’s book by Max Reinhardt for his production of Dantons Tod [Danton’s Death] (Freie Universität Berlin, n.d.).⁶ This project offers a wider audience the opportunity to participate in the creative process of a famous theatre director for the first time through a digital representation of the original director’s book. (Freie Universität Berlin, n.d., para. 1), Caplan (2015, p. 351), Estermann (2020, p. 33) and Illmayer (2017, slides 25–26) provide further overviews of various digital projects situated within a context of theatre studies.

This concise overview of selected projects highlights a clear research gap: no existing project systematically models data concerning on-stage activities. While current projects gather metadata about actors, venues, and theatrical materials, none integrates detailed staging elements alongside the production metadata. Consequently, data on stage sets, costumes, or props remains inaccessible and unrepresented.

(3) Performance vs. Production

Before introducing the technical framework and the data model, it is useful to clarify the distinction between a production and an individual performance. This distinction frames the analytical focus of the project.

The term performance refers to an ephemeral and non-repeatable event, shaped by the interaction of physically co-present performers (Gronau, 2012, p. 36). It constitutes the central object of theatre studies and, in contrast to literary works, is characterised by the heterogeneity of its artistic means, such as space, light, body, voice, and language (Brincken & Englhart, 2008, p. 7). A performance is tied exclusively to its moment of enactment, creating perceptual situations that exist only in this moment (Weiler & Roselt, 2017, p. 45).

By contrast, the term production refers to a conceptual framework that is developed during rehearsals and structures how scenic elements appear, interact, and convey meaning during performances (Brincken & Englhart, 2008, p. 107). A production is therefore a theatrical artwork or, from a semiotic perspective, a structurally organised arrangement of aesthetic signs (Balme, 2021, p. 99). As Weiler and Roselt (2017, p. 58) state, a production comprises all scenic components and their arrangement in relation to each other.

In addition to these two concepts, Doty (2013, para. 3) emphasises: “Live performance presents unique ontological challenges. We have not truly experienced a play unless we have seen a performance of it”.

This contextualization indicates that performances cannot be fully modelled unless they have been directly observed. A production, however, becomes visible through its performances. Probst and Pinto (2020, p. 173 footnote 25) acknowledge that productions are particularly challenging for data modelling because they are constituted by multiple performances in which almost all attributes, including venue, date, and even the performers involved, may vary from one performance to the next, with the director often being the only stable element.

Despite these challenges, this study deliberately focuses on productions rather than performances. This is because performance-specific characteristics such as audience perception, particularly that of theatre critics, and dynamic interactions cannot be meaningfully evaluated without having been physically present as part of the audience. Furthermore, the available sources rarely allow for a precise assignment of visual material (such as photographs in press reviews or theatre programmes) to a specific performance. Consequently, a performance cannot be identified when neither the provenance of photographs nor the specific performance attended by a theatre critic is documented.

In this research, the concept of the production is therefore understood as a concept that only becomes visible through performance but is nevertheless recognisable across several performances through recurring staging features. This notion builds on Balme’s (2021, p. 100) term of ‘constancy,’ referring to the assumption that performances of the same production share sufficient constancy to allow for intersubjective scholarly analysis of the production as a whole.

(4) Technical framework

This conceptual distinction that productions represent modellable, repeatable structures despite performance variability provides the foundation for both the technical implementation and data modelling approaches outlined in the following sections.

The decision to develop the data model within a dedicated Wikibase instance was primarily based on the aforementioned WikiProject Performing Arts, which provides an established modelling approach for theatre data, ensuring methodological continuity rather than creating a (new) isolated solution. No alternative databases were considered, as Wikibase enables seamless extension of existing Wikidata structures to accommodate staging characteristics like costumes and stage designs. The created Wikibase instance, hosted by the Trier Centre for Digital Humanities (TCDH), is accessible at https://mtm.uni-trier.de/ with the corresponding SPARQL endpoint at http://mtm.uni-trier.de/query/. Working on a separate Wikibase instance allowed experimental development in a protected environment to test the necessary new items and properties without disrupting Wikidata.

The data model is designed to support a highly granular representation of scenographic and production-related elements. This includes the ability to model costumes at component level, to specify attributes such as colour and contextual symbolism (e.g. the costume represents a uniform), and to link these components to roles and performers. The model is intended to capture similarly fine-grained structures for stage design, and multimedia elements, allowing individual objects to be decomposed into their constituent parts and semantically linked to their production.

In this way, the modelling framework aims not only to document productions at the level of traditional metadata (e.g., cast lists, venues, or dates) but additionally to encode what becomes visible on stage. This level of detail is essential for enabling advanced queries and comparative analyses. As additional productions are incorporated in future work, the model is expected to support scalable, interoperable, and analytically rich representations of theatrical performances.

Modelling decisions adhered to principles of findability, accessibility, interoperability, and reusability (FAIR) by assigning persistent, dereferenceable URIs to all entities and properties and linking them to corresponding Wikidata items where available (forschungsdaten.org, n.d.). Shared vocabularies from Wikidata enhanced findability and interoperability, while structured triples (except descriptive fields) ensured reusability through machine-actionable SPARQL queries.

The event-based structure draws conceptual parallels from CIDOC Conceptual Reference Model (CRM), abstracting ‘production’ as a collection of events (Bekiari et al., 2024). Thus, on the Wikibase instance, each production data object corresponds, for example, to an event (E5 Event), has a venue (E53 Place) or an actor, which in the case of CIDOC CRM describes every person involved (E39 Actor) (Bekiari et al., 2024).

(5) Dataset and sources

This research approach examines how production characteristics can be modelled using the Wikidata structure on a Wikibase instance using selected productions of Ferdinand von Schirach’s play Terror as case studies. The focus lies not on the quantity of modelled productions but on a structured and comprehensible data model that can adequately represent all important scenographic and contextual production characteristics of the performances.

Von Schirach’s play Terror premiered simultaneously on 3 October 2015 at the Schauspiel Frankfurt [Frankfurt Theatre] and the Deutsches Theater Berlin [German Theatre Berlin] (Gustav Kiepenheuer Bühnenvertriebs-GmbH, n.d.). During the 2016/17 theatre season, it became the most frequently performed play in Germany, with 36 productions (Dössel, 2018, para. 2). In mid-2018, the Nordwest-Zeitung [North-West newspaper] reported that “around 45,000 people worldwide have seen Terror so far” (Tschapke, 2018, author’s translation).

(6) Dataset description

At the current stage, the data set contains:

• 3 fully modelled productions of Terror

• 46 modelled individuals involved across directing, cast, design, etc.

• 100 properties, including over 30 newly created properties

• 15 costume objects

• References to a curated Zotero library with 17 referenced sources, all linked via source statements

While the dataset is intentionally limited to facilitate a controlled modelling process, it is designed to scale. As additional productions are incorporated, the analytical depth particularly for quantitative theatre analytics will expand substantially.

(7) Data modelling

The developed data model is based on and conceptually linked to the WikiProject Performing Arts. Existing entities were adopted and missing ones newly created. In the following paragraph, all existing and adopted items are identified by their Wikidata identifiers (Q- and P-IDs) prefixed with ‘WD’ for already existing items in Wikidata, to distinguish them from newly created items on the Wikibase instance, which have their own unique IDs.

Where a corresponding Wikidata item existed, the entities on this Wikibase instance were linked via the exact match property to enable federated SPARQL queries. Related data objects are bidirectionally linked to ensure consistent and traceable relations.

(7.6) Literary source integration

Finally, each production is connected to its literary source via the property based on (WD-P144). This entity of the written play contains statements about the author (WD-P50), title (WD-P1476), main subjects (WD-P921), narrative location (WD-P840), and characters (WD-P674) appearing in the play. This allows an analysis of the relationship between the literary source and stage adaptation. The roles within productions are in turn linked to the literary characters based on the literary figure (P87), enabling the examination of how these were realised on stage and whether, for instance, there were changes in role gender or reinterpretations differing from the literary original.

The result is a comprehensive and final data model (Figure 1) that enables a differentiated representation of productions, considering structural, scenographic, and receptive dimensions.

Figure 1

To assess the potential for integration, the WikiProject Performing Arts project was contacted. The feedback indicated that extending the data model directly within Wikidata is sensible only if a substantial volume of data is available and the model can be applied broadly across many productions. For cases focusing on qualitative, highly detailed descriptions of relatively few productions like the dramaturgical and scenographic features here, maintaining this data in a dedicated Wikibase instance remains preferable. General identification metadata can still be linked to Wikidata to ensure interoperability, balancing detailed case-specific documentation with scalability and sustainability.

(8) SPARQL queries

Building on the detailed data model described above, the following section demonstrates how structured information can be accessed through SPARQL queries, enabling researchers to explore the modelled productions.

To assist target groups in entering the query language, example queries are published on this Wikibase instance and categorised by subject areas (Wolter, 2025a). These include general queries providing overviews of the instance and productions; venue-related queries covering theatre buildings and non-traditional spaces (e.g. a courtroom); production setting queries addressing performance numbers, durations, and dates; people-related queries on individuals and their roles; gender-related analyses of casting; narrative setting queries on story locations; stage design queries on components and multimedia; costume queries on descriptions and components; and interactive element queries on voting types and visibility.

Each SPARQL query is provided with a link to open it directly in the SPARQL endpoint. In addition, some listed SPARQL queries include information about which variables users can exchange or modify to change the query semantics.

To enhance accessibility, the variables were carefully chosen so that each line clearly indicates what is being queried, and comments have been added to clarify the SPARQL code.

For instance, queries can identify productions where role genders differ from literary characters.⁹ Results reveal that in the Trier production, the defendant, defence counsel, and court officer were cast as female diverging from their male gender in the literary source. In the Berlin production, both the defence counsel and the judge were cast as female, whereas the original text depicts a male defence counsel and a male judge.

In addition to gender-related queries, detailed modelling of costumes allows users to retrieve how costumes are assembled across productions, including individual clothing item colours.¹⁰ The SPARQL query results, sorted by production and role, shows the components of each costume along with their respective colours. This reveals, for instance, that the judge’s costume in the Frankfurt production consists of a white necktie, a black judicial robe, a white shirt, grey trousers, and a grey lounge jacket.

(9) Challenges

Having demonstrated how the selected data were modelled within the data model and represented in the Wikibase instance, the following section addresses the challenges encountered so far in this research approach.

This project focuses on extracting production features from diverse sources and representing them in a digital-structured, machine- and human-readable form, suitable for querying and interoperable access. Accessing comprehensive sources, such as director’s prompt books, proved difficult, as many theatres no longer hold or archive them (Wolter, 2025b, p. 106). Illmayer (2017, slides 12–13) notes that, in addition to the ephemeral nature of performances, the multitude of different archival formats (especially audio and video recordings) pose a significant challenge. Moreover, theatres often accumulate their own collections of theatre materials, which are frequently unorganised and largely unknown.

Despite these challenges, the sources examined, such as print and online theatre reviews, proved to be more informative than initially anticipated. Texts and the frequently included images of performances enabled extraction of substantial information about productions and their features, though some gaps remained. For instance, the available sources do not allow reliable conclusions about the intentions of the directing team, such as how specific character traits were staged.

Further challenges may arise when modelling productions from performative genres or touring theatre productions, where venues and stage designs may change from one performance to another.

Beyond the archival and conceptual challenges, the development of an interoperable data model posed additional difficulties. Theatre productions consist of a variety of entities, including people, objects, roles, and spatial arrangements, which ideally all need to be represented in a form compatible with Wikidata, CIDOC CRM, and performing arts ontologies such as Swiss Performing Arts Ontology (Estermann & Schneeberger, 2017). However, not all scenographic elements have established equivalents in existing standards. Mapping fine-grained details such as costume components, stage design, or multimedia projections to existing Linked Open Data vocabularies revealed several gaps. While Wikidata offers a broad schema, its properties often lack the specificity required for detailed scenographic modelling. Conversely, CIDOC CRM and FRBRoo for instance provide conceptual depth but are not tailored to theatre-specific entities like roles, stage elements, or interactive audience components (Bekiari et al., 2024; Estermann & Schneeberger, 2017). Developing new properties in the Wikibase instance was therefore necessary, but these additions introduce potential friction when integrating the dataset into larger semantic ecosystems.

(10) Further possibilities

The Wikibase instance and its fine-grained data model open several concrete possibilities for extended analyses and research applications. Differences between the stage text and the director’s prompt book, such as deletions, additions, or textual changes, can be systematically encoded and queried to examine how productions interpret the original play. Similarly, simultaneous on-stage appearances of roles, as well as musical elements, can be represented as structured data objects linked to performers and scenes, allowing comparative analyses across productions.

Beyond the stage itself, backstage processes and production decisions can be documented in the model. Sketches, technical notes, and rehearsal materials can be integrated as facsimiles, capturing the evolution of stage design, costumes, and other production elements. This enables tracing of directional decisions, iterative alterations, and the development of staging concepts from the initial rehearsals and production meetings all the way to the premiere.

Finally, aggregating these data across multiple productions allows for the creation of director-specific production profiles or the identification of recurring stylistic patterns, such as the use of puppetry, multimedia components, or specific staging techniques.

(11) Reuse potential and target audiences

The developed data model offers multiple opportunities for reuse and engagement across diverse target groups both within and beyond theatre studies. By structuring productions, roles, stage design elements, costumes, and audience interaction elements in a digital, queryable, and interoperable format, the model enables a broad spectrum of analytical, and practical applications. Its design ensures that ephemeral theatrical practices can be preserved, analysed, and compared over time, supporting both immediate and long-term research needs.

(12) Conclusion

By digitally modelling the structural and scenographic dimensions of stage productions, this approach demonstrates how the ephemeral nature of theatre can be transformed into persistent, analysable, and interoperable data. Beyond its technical contribution, it offers new perspectives for how theatre as a live art form can be preserved, studied, and connected across disciplines through Linked Open Data, bridging the gap between the material and scenic dimensions of theatre.

Competing interests

The author has no competing interests to declare.

Author Contributions

Vivien Wolter: Conceptualization, Data curation, Investigation, Methodology, Writing – original draft.