Daejeon, July 27–31
How can a corpus reserved for a small circle of specialists be opened as a shared cultural object without sacrificing philological rigour? This question intersects with recent debates around small data, which stress the value of compact, well-described corpora for interpretation and model design. Within this context, this contribution presents eleven years of the Anthologia Graeca project (anthologiagraeca.org), an open-access, collaborative digital edition of the ancient collection of Greek epigrams known as the Greek Anthology, designed from the outset not as a closed critical edition but as a public, editable environment. Fragmentary epigrams, scholia, and editorial interventions are modelled as explicitly articulated entities within a flexible, extensible data model, turning the corpus into a web infrastructure where questions of access, interpretation, and computation can be investigated simultaneously and where annotation serves both computational experiments and broader cultural circulation.
Philological and literary studies have long traced the Anthology’s compilations and transmission (Aubreton 1968; Cameron 1993; Gutzwiller 1997), while media theory has emphasized infrastructural conditions and platforms (Méchoulan 2010; Doueihi 2011; Larrue / Vitali-Rosati 2019). The notion of editorialization – i.e. the set of operations that structure a text's circulation in digital spaces – bridges these traditions (Vitali-Rosati 2018, 2020). Within this framework, the project treats the collection as a dynamic object structured by acts of selection and commentary, series of epigrams, and recurring topoi (Vitali-Rosati et al. 2020; Mellet 2020; Verstraete 2024), and implements this view concretely in its data model, interfaces, and APIs so that these structures become visible and navigable beyond specialist communities.
The corpus comprises the modern Anthologia Graeca and consists of over 4,000 epigrams linked to thousands of related entities such as persons, places, and themes. For each epigram, the project gathers Greek text, translations, scholia, and links to images of the Palatinus 23 manuscript in a shared environment. Translations in multiple modern languages, notably French, English, Italian and Portuguese, support multilingual readership and teaching that reach audiences beyond professional classicists. The corpus is defined less by scale than by density: instead of pursuing large volumes, it invests in richly described, interoperable records designed for web circulation and integration into other tools, so that the Anthology can be reassembled in new reading paths and educational uses.
Technically, the infrastructure is implemented as a Python and Django-based web application with an openly documented REST API. Data circulate as JSON records conforming to a public schema and remain interoperable with related philological formats through import and export pipelines. Canonical identifiers and CTS-compatible URNs anchor records in a long-term citation infrastructure, while authority links to external resources (e.g. Wikidata) situate the edition within a broader linked-data ecosystem and facilitate reuse. Open data and stable identifiers enable a range of secondary uses and make the corpus an infrastructural testbed for different communities.
The project’s collaborative workflow is designed as a space of engagement and shared editorial responsibility rather than a pipeline reserved for a small group of experts. Through a web-based curation interface with versioned records, contributors can edit or create epigrams, revise texts and translations, add keywords and topoi, link scholia and manuscript images, align authority data, and validate results. These tasks are shared among students, research assistants, and senior scholars. Several hundred users consult the corpus each year and, should they wish to register, can act as editors, opening participation in philological knowledge production to a much broader range of contributors. Collaborations with schools and university courses integrate the edition into teaching, while thematic reading paths connect the live stack to non-specialist audiences without reducing its complexity, making the Anthology a shared laboratory for philology, digital literacy, and critical reflection on infrastructures. This layered participation broadens the editing community across educational levels, languages, and institutions.
Taken together, these infrastructures, workflows, and experiments frame the Anthologia Graeca project not as a single stable work but as an open ensemble shaped by media, infrastructures, and communities of practice. The project rethinks editing and annotation as situated engagement between readers, institutions, and technologies, and conceptualizes the Anthology as a dynamic, editorialized "imaginary" rather than a fixed text (Mellet 2020). Already used in ongoing work on explainable models of literary variation, this infrastructure offers a reusable model of small, annotated, collaborative editions in which preservation, linked-open-data circulation, and explainable AI are developed in tandem, making engagement with an ancient corpus a way for diverse audiences to question how contemporary digital cultures select, recombine, and transmit texts.