Daejeon, July 27–31
Museum exhibitions occupy a central position in contemporary artistic and curatorial discourses yet embody a fundamentally ephemeral form of cultural heritage. Once dismantled, exhibitions frequently vanish without leaving comprehensive traces, thereby compromising the conservation of essential cultural knowledge and the transmission of collective memory. As Bruce Altshuler (2008) and Vanessa Merminod (2012) have demonstrated, exhibitions are not only cultural artifacts but also sites of public, professional, and institutional engagement. Their inherently processual and relational nature resists traditional modes of documentation inherited from object-centered conservation and archival practices. We can consider that exhibitions, like performances, leave behind “performing remains” that challenge conventional archival logics (Parcollet et Szacka 2013; Hölling, Feldman et Magnin 2024). As a result, exhibitions often survive only through fragmentary records, limiting not only art historical inquiry but also broader digital humanities’ engagements with openGLAM, interpretation, and data modelling.
This paper argues that exhibition documentation constitutes a test case for contemporary digital conservation. It proposes that exhibitions can be understood as a form of small, complex, and difficult data whose preservation demands interpretive, situated, and ethically attentive modelling practices. Johanna Drucker (2021) reminds us that data are never raw; they are always “capta” (taken, not given) and thus reflect the values, assumptions, and interpretive frameworks of those who collect and structure them. Drawing on semantic web technologies, this research reframes documentation not as neutral recording but as an engaged act of knowledge modelling that actively shapes what becomes knowable, queryable, and transmissible about cultural events over time.
Three questions guide this inquiry: How can exhibitions be conceptualized as relational and performative cultural artifacts, and what does this shift demand of documentation practice? What are the epistemological limits of existing heritage ontologies when applied to exhibitions that deliberately destabilize authorship, identity, and temporal boundaries? And how can a documentary model accommodate not only uncertainty and absence, but the coexistence of multiple, equally legitimate situated perspectives on the same event? To address these questions, the paper proceeds as follows: with Feux pâles (capcMusée de Bordeaux, 1990-91) as a sustained case study, it first shows what composes an exhibition, surveys existing ontological frameworks and identifies their structural limits; then specify what an adequate documentary model in conservation would suggest.
The problem of exhibition preservation emerges from a structural mismatch between conventional heritage data models and the performative nature of exhibitions. Unlike discrete objects, exhibitions operate as assemblages of artworks, spatial configurations, technical infrastructures, institutional constraints, professional collaborations, discursive framings, and embodied experiences. Exhibitions unfold across time and often across multiple sites, producing meaning through relationships rather than through stable entities. Existing documentation systems, however, tend to reduce this complexity by privileging finished objects, named authorship, and fixed events, thereby marginalizing collective practices, informal labour, and experimental or non-dominant curatorial methodologies (Martini et Taramarcaz 2023). This reduction has significant implications for cultural memory, particularly for exhibitions emerging from feminist, Indigenous, queer, or community-based contexts whose modes of engagement frequently challenge institutional norms (Richter et Drabble 2015; Delhalle et Aurégan 2021).
Within digital humanities, the rise of linked data and semantic web technologies (Berners-Lee, Hendler et Lassila 2001; Bruseker et al. 2025) in museums has opened new possibilities for representing cultural heritage as interconnected networks rather than isolated records. Standards such as the CIDOC Conceptual Reference Model (CIDOC-CRM) developed by the International Committee of Museums (Doerr 2003; Doerr et Kritsotaki 2006), have enabled institutions to describe relationships among objects, actors, events, and places in machine-readable formats that support interoperability and computational analysis. Extensions and specialized ontologies have emerged responding to generic model limitations for exhibition documentation. LRMoo developed originally for editorial phenomena (RIva, Žumer et Aalberg 2022; Tietz, Bruns et Sack 2023), Onto-Exhibit addresses exhibition discursive dimensions (Rodríguez-Ortega 2018; Rodriguez Ortega et al. 2022; Rodríguez-Ortega 2024), the Art and Architectural Argumentation Ontology (AAAo) attempts to model difficult historical data without reduction (Takin solution et Swiss Art Research Infrastructure, 2025). We created Display ontology (Valentine, Château-Dutier et Renaudie 2024) within l’Ouvroir to focus on topological descriptions of installations. Despite this ontological richness, no model currently proposes an integrated approach specifically dedicated to exhibition documentation management. This paper contends that exhibitions expose the epistemological limits of existing heritage ontologies and invite a rethinking of semantic modelling as an interpretive and participatory practice rather than a purely technical solution.
This research adopts a pragmatic modelling approach (Ciula et al. 2023) in which the case study does not illustrate a pre-existing framework but actively generates the documentary requirements the model must meet. The central case is Feux pâles, organized by the agency © readymades appartiennent à tout le monde at the capcMusée d’art contemporain de Bordeaux from December 1990 to March 1991. The exhibition is an emblematic instance of what we term an exhibition-artwork: an institutional presentation that is simultaneously a fictional device. Visitors encountered what appeared to be a conventional thematic and chronological survey. Only on close inspection, and sometimes never, did the mechanisms of its construction become visible: the agency’s director was in fact the artist Philippe Thomas, and the entire curatorial apparatus was his fiction.
The methodological approach combines archival investigation of the exhibition’s material traces with theoretical engagement across conservation studies, museology, and digital humanities. It is grounded in the situated practice of conservation-restoration, in which sourcing information is not a preliminary step but a constitutive act: every observation is attributed to an identified professional, dated, and assigned an epistemic status : certain, uncertain, historically valid but no longer current, or deliberately left undeclared. A first attempt to document Feux pâles (Renaudie 2017, 2020) using a relational spreadsheet served as a diagnostic instrument: the inability to use the data, to model relations between heterogeneous elements, or represent competing interpretive accounts made the inadequacy of conventional tools immediately legible. This initial failure is the starting point of the inquiry. From there, this research works outward from the specific documentary problems posed by Feux pâles toward the identification of structural gaps in existing ontological frameworks, testing whether those frameworks can be extended or whether the case demands genuinely new modelling strategies.
Rather than modelling Feux pâles as a static entity, this research conceptualizes the exhibition as a dynamic and continuously reactivated network of actors, traces, interpretations, technical operations, institutional framings, and situated observations. The mapping against CIDOC-CRM, OntoExhibit, and the Display ontology revealed questions these frameworks were not designed to answer.
For exemple, after the exhibition closed, the displayed works were returned to their lenders, some works produced by Thomas entered the capcMusée's permanent collection while others dispersed into private and public holdings, and the catalogue is no longer in distribution. CIDOC-CRM can describe this dispersal event by event, but the mapping raises a question no framework currently answers: how does one document an exhibition as a coherent object whose material integrity no longer exists? After the show, Philippe Thomas created the installation Un cabinet d'amateur (1991) composed of twelve framed black-and-white photographs documenting the exhibition. It can only be adequately modelled through multiple instantiation, simultaneously an E22_Human-Made_Object and an E31_Document, artwork and meta-document at once. This double typing, which CIDOC-CRM formally permits, exposes how to model a documentary object whose status is not ambiguous but genuinely dual, and whose meaning shifts depending on whose account of the exhibition frames it.
The second concerns temporal boundedness. Feux pâles gave rise to subsequent events extending its existence beyond institutional closure, most notably its revisitation at the Mamco in Geneva in 2014 as L'Ombre du jaseur (after Feux pâles). Each iteration generates its own legitimate and irreducible account of what the work is. CIDOC-CRM's E52_Time-Span cannot hold non-linear or reactivable temporalities without forcing them into a sequential structure that distorts the work's logic. LRMoo offers a more adequate framework: its chain F1_Work → F2_Expression → F3_Manifestation → F5_Item allows each activation to be modelled as a distinct expression of the work, dated and situated, without being ontologically subordinated to a prior reference state (Delattre, 2026). Yet even LRMoo raises its own question here: it presupposes a stable conceptual work (F1) from which expressions derive, maintained through the distinction between F27_Work_Creation and F28_Expression_Creation. The relation between Feux pâles and L'Ombre du jaseur is not derivation from a stable source, it is activation: a continuation that simultaneously transforms and questions what the work is, and that retroactively redefines the F1 it ostensibly derives from. No existing framework seems to model this recursive instability.
The third concerns authorship, identity, and the multiplicity of situated accounts. The exhibition's fictional apparatus deliberately blurs the boundaries between artist, curator, institution, and collector, rendering any single authorized description a distortion of the work's logic. But the question goes further than unstable authorship: each actor involved constitutes Feux pâles differently, and these constitutions are not competing claims about the same stable object. For conservation, they are simultaneous and equally legitimate of what the exhibition is. For Claire Burrus, the artist's beneficiary, it is an artwork that can’t be reproduced. For the capcMusée, it is an institutional event that produced acquisitions and left archival traces. For Paul Bernard and the Mamco, it is a source that the 2014 exhibition L'Ombre du jaseur simultaneously honours and transforms. For the uninformed visitor who walked through the gallery in 1990, it was a conventional thematic exhibition, a reading the work itself solicited and depended upon. Following Goodman’s notion of “worldmaking” (Goodman 2011), this research argues that exhibition documentation must support the coexistence of multiple epistemic perspectives without collapsing them into a unified interpretation.
From this observation emerges a central methodological proposition: exhibition documentation does not primarily model exhibitions as objects, but rather the networks through which exhibitions persist, circulate, and become knowable over time. Documentation itself becomes part of the exhibition’s ongoing activation. In conservation practice, every statement is situated, attributed, dated, and associated with a degree of certainty, uncertainty, revision. The conservation study therefore does not stand outside the exhibition as neutral metadata production; it actively participates in the constitution of the documentary object. In the following practice of Collections as Data(Candela, 2024), this research consequently identifies a series of requirements for future documentary models dedicated to exhibitions. Such models should be capable of:
Feux pâles is not an exceptional case that exposes the limits of existing documentary models. It is a particularly legible instance of a broader condition: exhibitions whose meaning depends on the very instability, multiplicity, and opacity that conventional documentation seeks to eliminate. By working from the inside of this difficulty rather than around it, this research generates a precise map of what a more adequate documentary model for exhibitions would suggest. It demonstrates how a small, carefully structured dataset (when approached interpretively rather than exhaustively) can support nuanced engagement with complex cultural events, aligning with current digital humanities discussions around small data, annotation, and responsible computational practice (D’Ignazio et Klein 2020). This framework also raises a warning: not everything should be documented. Absence and opacity (Glissant 2009) are treated as legitimate epistemic stances, not failures to be remedied.
This paper positions semantic exhibition documentation as a form of engaged digital humanities practice that connects institutions, researchers, and communities through shared interpretive infrastructures. It demonstrates how modelling choices embody values, how documentation practices shape cultural memory, and how computational tools can be designed to support plurality, uncertainty, and ethical responsibility. By treating exhibitions as epistemic objects rather than logistical byproducts, this research contributes to broader DH conversations about memory, interpretation, and the role of digital technologies in mediating relationships between past events, present communities, and future forms of knowledge. Ultimately, exhibitions offer a productive lens through which to examine the promises and limits of semantic web technologies within the digital humanities, revealing that engagement is not only a matter of access or participation, but fundamentally a question of how cultural phenomena are modelled, whose expertise informs that modelling, and what forms of knowledge the resulting structures enable or foreclose.