Digital access to collections does not guarantee their intellectual accessibility. In art institutions, one of the major challenges remains the semantic gap—that is, the discrepancy between the specialized vocabulary used by information professionals and the natural language used by the public in their searches. In response to this issue, citizen contribution, particularly through social tagging and collaborative transcription, emerges as a means of democratizing access to knowledge and diversifying entry points to collections.
Citizen contribution can be understood as a form of participatory documentary mediation, situated at the crossroads of description, interpretation and use, and involving a partial redistribution of documentary authority. It challenges traditional hierarchical models of knowledge production, historically based on expert authority. By drawing on users’ experiential knowledge, these initiatives encourage a plurality of interpretations and contribute to redefining the role of archival institutions as relational spaces, where knowledge is constructed collectively, through exchange and shared presence.
From authority to collaboration: the pivotal moment
Practices, dynamics and issues in citizen participation
The history of citizen contribution in heritage institutions can be analyzed in four phases: inception, structuring, expansion and technological hybridization (Wyman et al. 1; Chae & Kim; Smithsonian Institution; BrodeFrank 47). This classification describes a technical evolution, but more fundamentally, a gradual shift in documentary authority regimes. Between 2005 and 2008, the rise of folksonomy marked an initial break with models strictly based on controlled documentary vocabularies (Wyman et al. 1; Trant, Tagging, Folksonomy and Art Museums 3–4; West 17). The Cataloguing by Crowd forum (2005) demonstrates that databases designed by and for experts remain largely opaque to the general public, thereby limiting the uptake of digital collections (Wyman et al. 1). The issue is no longer merely one of access, but of the legibility and recognition of everyday ways of naming and understanding art. The pioneering steve.museum project, launched in 2006 by a consortium of North American museums, demonstrates empirically that non-specialists can enrich the metadata of artworks in meaningful ways that are complementary to, and sometimes absent from, expert systems (Trant, Tagging, Folksonomy and Art Museums 37). As early as April 2005, the Cleveland Museum of Art was experimenting with a prototype entitled Help others find me, based on the social altruism of users (Wyman et al. 2). These early initiatives shift the descriptive function of the artwork: from an act of authority to a relational and dialogical act.
Structuring and gamification of practices
Between 2008 and 2015, institutions sought to channel the sometimes chaotic nature of free-form tags (spelling mistakes, redundancies, inaccuracies) without discouraging participation (Chae & Kim; West 17–18). This phase is characterized by a central tension: how to structure contributions without neutralizing their expressive richness or compromising their integration into the documentary record? In 2010, the Gyeonggi Museum of Modern Art introduced a multi-faceted tagging system (theme, context, emotion), enabling contributions to be guided whilst preserving the diversity of perspectives (Chae & Kim). This gradual structuring aims less to discipline audiences than to translate their contributions into a language that can be processed by documentary systems. At the same time, gamification is becoming a key driver of engagement. The ARTigo project, developed by the University of Munich, transforms image tagging into a “game with a purpose,” mobilizing over 43,000 participants and generating millions of tags (Schneider 720). Here, the game acts as an invisible mediating mechanism, where the documentary effort is absorbed by the playful experience, making the contribution accessible and sustainable. Initiatives such as the Brooklyn Museum Posse also illustrate the potential of these mechanisms to support regular and sustained participation. Launched in 2008, this project relied on playful dynamics and a points system to encourage users to enrich the collections, thereby generating over 58,000 tags in ten months before incorporating a peer-to-peer quality control mechanism via the game Freeze Tag (West 23–24; Yuan Li 4).
Democratization and mass engagement
Since 2013, public contribution projects have been characterized by a growing commitment to transparency and large-scale inclusion (Smithsonian Institution; BrodeFrank 149; Library of Congress). Participation is no longer viewed as a prototype or an add-on, but as a structural component of digital strategies. The Smithsonian Transcription Center mobilizes volunteers, known as volunpeers, to transcribe historical documents, making thousands of pages of archives available in full text (Smithsonian Institution). Meanwhile, the Library of Congress uses Flickr to share its copyright-free collections and enable users to enrich the data. This data is then incorporated into the descriptive records of its online catalogue (Library of Congress). These initiatives reposition institutions as platforms for collaboration, rather than mere producers of knowledge, and reinforce their role as relational infrastructures.
Towards a human—AI hybrid
Recent developments are part of a trend towards hybrid workflows, combining artificial intelligence with human judgment. Initiatives such as Tag Along with Adler or the hackathons at the Metropolitan Museum of Art explore the use of AI to suggest descriptive terms, which are then validated by human contributors (BrodeFrank 47; Lih). AI does not replace human interpretation: it acts as a catalyst, accelerating the generation of proposed terms and highlighting discrepancies, disagreements and cultural nuances. This hybridization highlights a reconfiguration of documentary work, where the machine assists, but where meaning remains collectively negotiated (BrodeFrank 53).
Documentary strengths and challenges
Research from steve.museum shows that 86% of tags produced by the public do not appear in traditional museum documentation (Trant, Tagging, Folksonomy and Art Museums 37). Citizens describe what the artwork represents (aboutness) rather than what it is physically (isness), thereby multiplying access points. This semantic richness broadens research possibilities, but also redefines what is deemed relevant or worthy of mention within the institutional documentary space. However, the “power user” bias—where a minority is responsible for a disproportionate share of contributions—raises issues of representation (Trant, Tagging, Folksonomy and Art Museums 3; BrodeFrank 87). These contributions must, however, be analyzed with caution. They reflect cultural, linguistic and social factors specific to a given time and place, which may lead to inequalities in the representation of works. Without strategies for active inclusion and diversification of the contributing public, participation risks reproducing certain asymmetries already present in institutional collections.
Conclusion
Citizen’s contribution transforms art institutions into a space for polyphonic dialogue, where professional expertise coexists with collective intelligence. By pooling their infrastructure, combining human validation with artificial intelligence, and supporting audiences through educational initiatives, art institutions may not only reduce the semantic gap, but also turn metadata into a living, evolving common good that reflects the plurality of perspectives on art.
References
BrodeFrank, Jessica. “Crowdsourcing Metadata in Museums: Expanding Descriptions, Access, Transparency, and Experience.” Perspectives on Data, ed. Emily Lew Fry and Erin Canning, Art Institute of Chicago, 2022, pp. 43–54.
Chae, Gunho, and Jungwha Kim. “Can Social Tagging Be a Tool to Reduce the Semantic Gap between Curators and Audiences?” Museums and the Web 2011: Proceedings, Archives & Museum Informatics, 2011, https://museumsandtheweb.com/mw2011/papers/can_social_tagging_be_a_tool_to_reduce_the_sem.
Kohle, Hubertus. “ARTigo: social image tagging for works of art.” Art and Measurement: Art History and Quantitative Methods, ed. Béatrice Joyeux-Prunel, Éditions de l’École normale supérieure, 2010, pp. 153–162, https://books.openedition.org/editionsulm/8657.
Library of Congress. “Frequently Asked Questions—Photographs on Flickr from the Library of Congress Collections.” Research Guides at Library of Congress, n.d., https://guides.loc.gov/flickr.
Lih, Andrew. “Combining AI and Human Judgment to Build Knowledge about Art on a Global Scale.” The Metropolitan Museum of Art, 4 March 2019, https://www.metmuseum.org/perspectives/wikipedia-art-and-ai.
Schneider, Stefanie. 2025. “ARTigo: Lessons From Social Image Tagging in an Art— Historical Game With a Purpose.” IEEE Transactions on Games, vol. 17, no. 3, 2025, pp. 720–728.