Conference Report: Workshop Objekt - Text - Bild, Berlin, April 28 2025

1 Workshop Objekt – Text – Bild NFDI4Objects Community Cluster „Objects as information carriers” Berlin, April 28 2025 NFDI4Objects Workshop: Objects as information carriers – summary The workshop of the Community Cluster ‘Objects as Information Carriers’ took place on 28 April 2025 at the Archäologisches Zentrum Berlin and online and brought together experts from archaeology, art history, Egyptology, numismatics and digital humanities, including from the NFDI consortia Text+ and 4Memory. The aim of this event, which followed an initial workshop in October 2024 (Fig. 1), was to discuss current projects and challenges in the digital indexing and modelling of inscription carriers, particularly with regard to standard data, interoperable metadata models and ontologies, as well as the creation of synergetic strategies in various projects and consortia. The DOIs of the contribution presentations are listed at the end of the synopses. Fig. 1: Programme of the 1st Workshop of the Community Clusters on 22.10.2024 Part I: Project presentations and central themes Anne Herzberg-Beiersdorf presented her project on the digital recording of private monuments from Memphis (New Kingdom period, ca. 1550-1070 BC). Over 1700 objects, 2409 persons and 992 Egyptian official and functional titles had already been recorded in a PostgreSQL database as part of her dissertation project and put in relation to each other. The methodological framework of social network analysis was used to prepare and analyse the data. The database records various metadata categories, such as information on material, dating, provenance and location, using standardisation data sources (e.g. MET, THOT and Trismegistos, TopBib, Geonames and Wikidata). A central challenge is the standardisation of the location and position of the inscriptions in their respective superordinate context (e.g. the carrier medium, 2 archaeological/epigraphic context), as no standardised specialist vocabularies for archaeological contexts exist to date. https://doi.org/10.5281/zenodo.17020596 Alexander Ilin-Tomich provided a brief insight into his prosopographical project on Egypt in the Middle and Early New Kingdoms. The database of 53,198 individuals is published both as a web-based MySQL database and as a graph database. The data model requires in particular the links between people, iconography, language, history and archaeology, with the speaker describing a gap between field archaeological, museological, historical, pictorial and text-based disciplinary specialisation as emblematic also of the workshop theme. The modelling of the relationship between objects and inscriptions is based on an ontology referenced on CIDOC CRM, whereby the distinction between physical objects, virtual objects and texts remains problematic. In particular, multi-part, non-linear texts (e.g. palimpsests) and objects consisting of several finds or museum pieces illustrate the erroneous conclusions caused by heterogeneous ontological classifications (e.g. E22 human-made-object vs. E24 human-made-thing) and different definitions of object and inscription. https://doi.org/10.5281/zenodo.17021378 Daniel Werning presented the modelling of artefacts and texts in the Thesaurus Linguae Aegyptiae (TLA). The data model of the TLA differentiates between texts (BTSText) and objects (BTSObject), whereby shortcomings were also described that can already be solved to varying degrees of satisfaction. Particular challenges result for the precise integration of object data from fragmented text carriers, multi-part and non-linear texts as well as pictorial elements. An exemplary mapping to the ontology of the FAIR Epigraphy Project led to a largely accurate fit of the modelling there, in which mainly only the transfer of individual classes used there to the Egyptological requirements left certain ambiguities. Various detailed solutions were proposed for the self-critically described shortcomings in the integration of text-object references in the TLA, for example through segmentation and annotations. https://doi.org/10.5281/zenodo.17021761 In his contribution, historian Philipp Schneider focussed on digital heraldry. The coats of arms of the later Middle Ages and the early modern period are compositions made up of various layers, including the conceptual arrangement of individual pictorial elements, colours and shapes. The DigitalHeraldry project (headed by Torsten Hiltmann) has created a knowledge graph with four (sub-)ontologies (Objects, Representations, Entities, Heraldic Elements) for heraldry from RDF-indexed data from a local heraldry database. The ontology models the complex relationships between coats of arms, their components and contexts, which are crucial for the adequate interpretation of the coats of arms. The challenge lies in the fine-grained analysis of the heraldic elements as well as in the integration of pictorial data (e.g. in IIIF) and multilingualism. In the follow-up project HERALDic Identity in Context, the data model and the ontology are to be made open access. https://doi.org/10.5281/zenodo.17022254 Teil II: Impulse lectures and practical modelling The model presented by Sarah Wagner sees the inscription of objects as part of the digital object biography and serves to integrate heterogeneous and distributed object information. The object is considered in the context of events, such as its creation, its 3 use, a museum or collection context, its modification or research activities, each of which can be linked to actors, time and place and must always be documented with sources. The object biography is modelled using the N4O Objects Ontology, whose top-level ontology is CIDOC CRM version 7.1.3. Inscriptions are currently recorded in the digital object biography in two contexts: when the object is created or produced (E12 Production) and when it is modified (E11 Modification), for example when inscriptions are added, restored or otherwise altered. When recording an inscription (E34 Inscription), its transcription (E31 Document) and language (E56 Language) can be documented. The inscription, in turn, is linked to a physical feature (E25 Humanmade Feature) by which it is carried (P128 is carried by) and on which it can be located (P59 has section). Particularly in the case of objects that bear several inscriptions, the question arises as to the level of detail of the recording (case study Behaim globe from 1492/94 in the German National Museum Nuremberg, which bears over 2000 place names). In the course of the discussion, the question was raised as to whether inscriptions should not also play a role in the context of the use of an object, for example when they are read aloud by certain actors. https://doi.org/10.5281/zenodo.15632274 Anja Gerber shared modelling approaches for integrating heterogeneous archaeological data into the NFDI4Objects Knowledge Graph. Multidisciplinary application examples from the N4O Task Areas 1-4 are required for this task, which is largely driven by the ‘N4O Object Metadata Set & Object Ontology’ TWG. A path was outlined that uses LIDO as a metadata schema to describe object data based on LIDO (XML) for the N4O ObjectMetaDataSet. By referencing CIDOC-CRM 7.1.3, among others, an RDF-capable representation is then created in the N4O Object Ontology. A key challenge is to map the different data formats interoperably and without loss of information. Approaches were presented for modelling complex objects, multi-part and non-linear texts as well as pictorial elements. It is particularly important to determine which relationships between objects, texts and images should be expressed with which properties. The approach of the minimum data set recommendation for museums and collections used as a starting point for the metadata model is not sufficient to adequately represent more complex relationships between texts and objects. In close cooperation between NFDI4Objects, the Minimal Dataset Working Group and the German LIDO AG, the needs of the consortium will be incorporated into the further development of the minimal dataset and an archaeological application profile for LIDO. https://doi.org/10.5281/zenodo.17023156 Christoph Klose used a coin type from the late 1st century BC to present the current situation and shortcomings in the recording and visualisation of text-object relationships in numismatic research data. Despite the diverse use of standard data for persons, corporate bodies, geographical locations and other numismatic concepts, modelling the linking of pictorial and textual information remains a challenge. When exporting in the IKMK’s LIDO-XML data format, the legend and image description are mapped in the ‘Inscription’ and ‘Inscription Description’ elements, which can result in insufficient separation of the textual and pictorial object properties, requiring alternative ontological referencing, for which suggestions have been made: A separation of inscription information into the Lido concepts of ‘Inscription Transcription’ (for the reproduction of textual information) and ‘Inscription Description’ (for the description of formalities such as direction or characteristics such as engraved inscriptions). https://doi.org/10.5281/zenodo.17023338 4 Key aspects of Florian Thiery’s contribution, of which due to unfortunate time constraints only the presentation could be made available and briefly summarised, concern the mapping of special application ontologies to CIDOC CRM and the use of the TEI Epidoc metadata schema widely used in (classical) epigraphy on the basis of two object resources: Celtic Ogham stones and Roman stamped pottery (see in particular slides 2, 10, 14, 15, 25 and 27 of the presentation linked below). Fig. 2: Illustration of the integration of different ontologies in N4O knowledge graph. Slide 2 from Florian Thiery's presentation, ©Florian Thiery CC BY 4.0 A key point of the contribution is to emphasise the interoperability of different metadata schemas and ontologies and their integration into the N4O Knowledge Graph. In addition to the examples frequently mentioned in the workshop (such as LIDO and CIDOC CRM), this also applies to Wikibase instances and Factgrid (Fig. 2). https://zenodo.org/records/15310724 Babette Schnitzlein focussed on the metadata of cuneiform artefacts. The recording of museum numbers including joins, text editions, script (levels) and classification is essential in order to make textual evidence usable in digital collections. The standardised documentation in the ‘Reading the Library of Assurbanipal’ project (LMU and BM London) for plates and plate fragments with colophons offers good approaches, for example through the use of harmonised localisations, such as in the classification. The latter is (now revised several times) also used in the electronic Babylonian Library (LMU and Bavarian Academy of Sciences and Humanities). The discussion emphasised the need to integrate such standardisations with stable identifiers as authority data in other databases and to improve data quality through clear guidelines. https://doi.org/10.5281/zenodo.17023440 Overarching topics, questions and future steps Key questions regarding the customisation of the CIDOC CRM classes and properties were discussed in order to adequately map the diverse object properties. There is a need for low-level and top-level ontologies as well as for the modelling of multi-part 5 objects, e.g. funerary steles located in different museums. The modelling of serial inscriptions and multi-part text carriers was repeatedly described as a challenge. For the next steps, the collection of specific mappings of metadata to standards such as CIDOC CRM, Wikidata, Factgrid, LIDO XML and TEI Epidoc was agreed. The aim is to drive forward the integration of selected object texts into the shared database and improve interoperability between the various data sources. Best practice examples are also to be developed in order to standardise and simplify the modelling of complex inscription carriers in digital collections. Christoph Klose