TypeDB helps advance research in digital archives
A new paper, published by Kenneth Thibodeau (retired from National Archives and Records Administration), Alex Richmond (Bank of Canada), and Mario Beauchamp (Carlton University), shows how TypeDB has enabled a multi-year project that seeks to develop better digital archival research tools.
Background
Archival theory on its own focuses on the organization and functions of current and historical archives. Archival research theory (ART) enriches this by also including the organization and functions of research interactions with archives.
A researcher in digital archives should be able to access records without bias or any interference with the researcher’s objectives and research lenses. An archive that conforms to the goals set out in archival research theory “should not limit what researchers can find in archives or constrain the level of specificity they need in characterizing it.”
This is an ambitious goal, and incurs inherent complexity. Firstly, it requires preserving the base facts, information about interpretation contexts, and layered data about the creation of the archive as well as internal and external relations. And secondly, it requires providing information responsively to specific interests.
Choosing TypeDB
The researchers chose TypeDB for its expressivity and ability to model complex, layered, and contextualized data.
To quote: “The PERA model provides expressivity and coherence for modeling complex domains as well as for relating data obtained from a variety of sources,” going on to say, “TypeDB facilitates articulating schemas that more fully and faithfully reflect domain knowledge than other alternatives. Its expressive schema model and intuitive query language make it ideal for representing the interconnected, evolving knowledge…”
The model
The paper implements a granular, highly contextual schema in TypeDB that follows ideas from semiotics. Semiotics studies how meaning is determined. By explicitly modeling a high-level, abstract schema of semiotic objects, meaning can be interpreted through the database itself.
The ART schema provides ways for researchers to separately specify what something in or about an archive means to them, from what it meant to agents who created the information objects, as well as to recognize differences in meanings among interactants. Concretely, the schema implements its archival research objectives with nested relations like events, actions, and reactions, allowing researchers to capture many dimensions of meaning, context, and causality.
The type-theoretic approach TypeDB offers, where data meaning can be steered by the roles that are played, is key. For example – the core action relation is an n-ary hyper relation: it has 7 roles – active, passive, stimulus, response, enabler, condition and outcome. These give context to things like resources, agents, patients, events, or reactions (all themselves relations!). Even more, the polymorphic application of these roles are extended to *either* entities or other relations, a capability few other databases offer.
The ART schema also leverages subtyping, allowing specifying different types of semiotic resources, conditions (control, norm, and environmental controls, each with their own subtypes), etc. The paper proposes an expansive abstract data model, specifying concrete layers of types below it. This lends itself to customization and implementation according to different archives’ needs.
Here’s a visual representation of an part of the ART schema from the paper:
The goals
An archive following ART will have enhanced semantics to enable maximizing the potential of the data, as well as for sharing data among researchers and programs. TypeDB has enabled these researchers to develop a “comprehensive and durable conceptual foundation,” in the shape of a TypeDB schema.
Check out their paper here.
Our perspective
As CTO of TypeDB, it’s incredible to me to hear about all the novel and interesting ways TypeDB is being used to solve hard problems. These success stories are what motivate us to keep building better, stronger versions of the product.
Whether you’re pushing the limits of your field, building applications over structured knowledge layers, instrumenting AI agents with long term context and memory… we want to hear from you!
And as always, feel free to come and chat with our community on Discord or email us at community@typedb.com.