Session 4

Data structures

Author

Francesco Beretta

Published

6 October 2026

Modified

2 September 2026

Beretta Taught by Francesco Beretta.

In this session

Conceptual modelling: what a data model is, before any database exists
A showcase of relational and graph structures
Two modelling exercises
Reflection on domains and disciplinary diversity
Dublin Core and METS

Conceptual modelling

Following (Flanders and Jannidis 2018): a data model is an argument about what exists in your material and how it relates. It is made before the technology and it constrains everything afterwards.

Entities, attributes, relations
What a model asserts, and what it makes unsayable
Why the first model is always wrong, and why that is fine

Showcase: relational and graph

Two ways of holding the same material, side by side:

Relational — tables, keys, joins; strong on constraints and counting
Graph — nodes and edges; strong on connection and traversal

We are not choosing a winner. We are looking at what each makes easy and what each makes hard.

Exercise 1: articles and authors

Build a data model for articles and their authors — the material from your own Zotero library.

What is an author? A string? A person? A person with an identifier?
What happens to co-authorship, to editors, to institutional authors, to pseudonyms?
Where does a journal sit in the model — attribute or entity?

Draw it. Paper is fine; the Lab introduces draw.io and QuickDBD the same week.

Exercise 2: artwork

Now model a work of art — an object with a maker, a material, a date, a location, and a history of all of those.

How do you model a date that is “around 1480”?
How do you model an attribution that changed in 1962?
Where does the object end and the record about it begin?

Reflection: diversity of domains in the humanities

Different disciplines model different things, and the differences are not superficial:

Prosopography — people, roles, relations, uncertain identity across sources
Materiality — objects, substances, condition, provenance, physical description

Bring your own field into this. What does your material insist on that the models above cannot express?

Standards: Dublin Core and METS

Dublin Core — fifteen elements, deliberately minimal; what that minimalism buys and what it costs
METS — structural metadata for complex digital objects; describing not the content but its arrangement

In parallel in the Lab

The DH Lab turns this into practice: replacing Excel with SQLite.

Reading for Session 5

Beretta, Francesco. 2024. “Conceptualising Information Production in the Context of the SDHSS Ontology Ecosystem.” Methodos. Savoirs et Textes 24. https://doi.org/10.4000/12xqn

References

Flanders, Julia, and Fotis Jannidis. 2018. ‘Data Modeling in a Digital Humanities Context: An Introduction’. In The Shape of Data in Digital Humanities: Modeling Texts and Text-based Resources, edited by Julia Flanders and Fotis Jannidis. Routledge. https://doi.org/10.4324/9781315552941.