The poster presents an approach to integrate temple decoration into a single model with a graph database, using the temple of Hatshepsut at Deir el-Bahari as a case study. By linking data from the decoration programme to the layout of the temple, patterns in the spatial distribution of elements within the decoration program can be examined. The poster focuses on the methodology and presents a use case from the ongoing project where such an approach contributes to the interpretation of ritual space.

The spatial layout of Egyptian temples has been subject to numerous studies (examples include Arnold 1962, Białostocka 2016, Baumann 2017, Ullmann 2002); layouts of New Kingdom (ca. 1539-1077 BCE) Temples of Millions of Years followed firmly established, structural patterns. The temple of Hatshepsut was first published in excavation records by Naville (1895-1908), with research ongoing to the present day (e.g. Arnold 2005, Barwik 2021, Barwik 2010, Białostocka 2016, Białostocka 2010, Kapiec 2021, Karkowski 2003, Keller 2005, Roth 2005, Stupko-Lubczyńska 2016, Szafrański 2010). The framework applied to interpret the observations made in the temple is Multimodal Communication which has been successfully used in Egyptian contexts (Kutscher / Serova 2025); an XML-based annotation tool for Egyptian scenes with this framework as its base has also been developed (Döhl et al. 2023). Data modelling in Egyptology often follows general trends of database development (Dietrich 2025). Notable work employing graph databases includes Herzberg (2021), and Dulíková / Barta (2020), with methodical overviews for Egyptological network data in Stefanović (2024) and Chollier (2019). Beyond Egyptology, the possibilities of working with in graph databases in the humanities has been widely explored and published (e.g. Ahnert et al. 2020, Kuczera 2019, Schmidt et al. 2022).

As graph databases are particularly well suited to model semantic links (Kuczera 2019: 14), they can be used to build knowledge bases for cultural heritage. For scenes in the graph database presented here, existing descriptions from Porter et al. 1972 were reused to create basic semantic scene annotations. The collected data includes annotated depictions – linked to their respective spatial localisation, most commonly the scenic context – as well as their underlying concepts. This database was realised in neo4j with a custom python-based frontend and has already been used for interpreting scenes on a localised level in previous work (Dietrich 2023). Its modelling strategy enables researchers to record different parts of the decoration programme, i.e. texts, scenes, objects and statues, in a single database with semantic links and to compare depictions in their context for different parts of the temple. Spatial connections are modelled as relative values as the data model incorporates a mapping to the concrete placement in relation to other elements in the temple. All decoration elements are linked to one or more entries in a thesaurus and then connected semantically to each other. By examining emerging local and global connections in the database, outliers and constants in the data can be observed. Using the framework of multimodal communication, these observations facilitate a better understanding of how ritual space in this specific temple and its context was constructed. Through a mapping of the project specific data model to the CIDOC-CRM (Bekiari et al. 2025), the data from this project can be incorporated into other systems describing cultural heritage.

To illustrate results obtained with this method, the poster presents the use case of interpreting depictions of the god Thoth with the king. The locations where the god and the king appear together align almost perfectly with areas of importance to the royal cult. This includes the birth portico and the upper terrace rooms with the ritual spaces dedicated to Hatshepsut and her ancestors. Scenes with the king and Thoth, the latter structurally taking precedence over other gods, are thus closely related to ideas of the royal succession. This observation adds another piece of evidence for linking the god to the royal family, rather than his typical domains, which has been observed from other sources (García Fernández 2024). It also aligns with the importance of decoration programs along routes through the Egyptian New Kingdom temples (Revez / Brand 2015: 253-254). Structural queries like this can easily be modelled and executed in a graph database. Their results, as in this case, highlight semantic links that could not have been observed in traditional work focused on the global level (for comparison: Stupko-Lubczyńska 2016: 283, 317-320).

Using a semantically linked graph database in use cases like this one therefore enables researchers to better understand the way ritual space and its meaning was constructed. This approach combines traditional scholarship and its output with the capabilities of digital data modelling. Further work, however, is needed to assess the accuracy and level of detail present in previous scene descriptions and to expand the data set, especially text and reconstructed statue placements. The project database and associated software will be made available under an open license after the project is concluded.