DNKB Structured Data Guide
A complete reference guide explaining how metadata, connections, and records work inside the Digital Narrative Knowledge Base. Learn how to read, build, and contribute clean, queryable knowledge.
1. Data Guide Overview
The Digital Narrative Knowledge Base (DNKB) documents digital narrative objects—such as creative works, critical writing, authors, events, and technologies.
To make this knowledge archive useful for computers and humans alike, DNKB separates its content into two formats:
description Readable Prose / Narratives
Used for descriptive text, interpretation, historical context, and critical analysis. This is meant for humans to read.
database Structured Data / Records
Individual, atomic facts. Because they follow rigid guidelines, this data can be searched, sorted, filtered, and linked to other databases.
Use this guide when:
- You are learning to add or edit records for the first time.
- You need a refresher on the difference between Item and String value types.
- You want to review common mistakes before publishing your revisions.
2. Core Concepts
DNKB uses a triple-style graph structure where all data is composed of **Items**, **Properties**, and **Values**. These combine to form **Statements**.
1. Items
An **Item** represents a distinct entity, idea, or concept. Every work, person, genre, event, or language is its own Item and has a unique ID (like Q130).
2. Properties
A **Property** describes the relationship. It's the field name or attribute that describes how an item relates to its values. Examples include author, genre, and publication date.
3. Values
A **Value** is the actual data content recorded for a property. If the property is author, the value is the linked Item for that person. If the property is a date, it is the year.
3. Value Types
Not all information is structured the same way. DNKB supports several value types to ensure facts remain structured and consistent.
Links a statement directly to another entity in the database (e.g., connecting a work to a publisher or platform).
Plain text. Use only when the value is just text and does not represent an entity that needs its own page or properties.
A structured calendar date or year. Used to track timelines, historical releases, and creation times.
A fully qualified hyperlink pointing to external archives, projects, databases, or live websites.
A specific ID from another authority catalog (e.g., Wikidata ID, Quora Topic ID, or ORCID).
Rule of Thumb: Items vs. Strings
Ask yourself: Is this value a concept or entity that has (or could have) its own attributes, relations, or page?
- If yes (e.g., an author, a database platform, a primary publishing house), make or find an Item and link to it.
- If no (e.g., a simple comment, an arbitrary quote, a spelling variation), use a String / Text value.
4. Qualifiers & Sources
Statements can express simple facts, but real history and metadata are often more complex. DNKB supports two ways to refine and back up statements:
tune Qualifiers
Qualifiers add context or details to a statement without changing the primary statement itself.
For example, to state that a work was distributed through a store only until 2004, you would add the qualifier:
└─ Qualifier: end time → 2004
local_library Sources & References
Sources cite where your statement's data comes from. Adding references makes the knowledge base verifiable and trustworthy.
For instance, if you document that a specific author won an award, back it up by referencing the official announcement URL:
└─ Reference: reference URL → http://eliterature.org/...
5. DNKB Ontology
DNKB uses a local structure (an "ontology") to organize what kind of concepts are recorded. This helps separate the primary subjects of digital narrative study from helper data and secondary supporting records.
grade Primary Research Objects
These are the central focus points of our research and databases.
- Creative Works: Interactive stories, hypertext literature, video games, net art, digital poems.
- Critical Writings: Academic papers, critiques, books, and essays discussing digital narrative.
- People / Authors: The writers, developers, artists, and scholars behind the works.
settings_accessibility Secondary Supporting Records
Entities that provide metadata and institutional context for primary objects.
- Human: Author or researcher profile pages (inheriting birth/death/citizenship data fields).
- Publisher: Publishing houses, platforms, or organizations.
- CELL Project: Associated sub-projects from the Consortium for Electronic Literature.
- Academic Journal: Periodicals where writings are published.
- Event: Contextual events like conferences, festivals, and award ceremonies.
lists Helper & Category Items
Controlled-vocabulary values used to describe and link primary objects.
- Platforms: Microsoft Windows, iOS, web browser, Commodore 64. What counts as a platform.
- Software engines: Flash, HTML, Twine, Unity.
- Genres: Hypertext fiction, interactive fiction, generative poetry.
- Languages & Countries: Natural languages and country items.
6. DNKB & Wikidata Alignment
DNKB does not exist in isolation. Contributors are **strictly required to align local items with Wikidata** using the `Wikidata ID` external identifier. This alignment must be maintained even in cases where it feels inconvenient, requires additional lookup, or demands extra modeling effort.
High-Trust Curated Database vs. Public Wikidata
While we link directly to Wikidata to connect with the global semantic web, DNKB has a completely different curation model:
1. Researcher-Only Maintenance: Unlike Wikidata, which is editable by any anonymous or public user and can suffer from inconsistent schemas and unverified edits, DNKB is maintained and edited exclusively by qualified researchers. This ensures a highly accurate, peer-reviewed, and academic dataset.
2. Specialized Scope & Invariants: DNKB maintains its own specialized ontology and review processes tailored specifically for electronic literature and digital narrative studies. Alignment allows us to anchor our highly verified local records to the broader web.
7. Common Mistakes to Avoid
Contributing clean metadata requires keeping format boundaries clear. Here are the most common modeling issues to watch out for:
cancel Incorrect Practice
- Typing the author name as a String:
author: "Shelley Jackson". This prevents links! - Writing a full critical summary inside a text field.
- Creating a duplicate item because you did not search the database first.
- Adding values with the wrong type (like linking a URL inside a Date property).
check_circle Correct Practice
- Link the Item:
author: Q1202 (Shelley Jackson). - Put descriptions in narrative prose sections on the main page.
- Always perform a search before creating a new entity.
- Match property constraints strictly (e.g. use proper date syntax for dates).
8. Quick Reference
Use this table to quickly check definitions when editing or contributing records:
| Concept | Short Definition | Example |
|---|---|---|
| Item | An entity, page, or concept in the database. | Q7007 (Creative Work) |
| Property | The relationship attribute or descriptor. | P6 (author) |
| Value | The target data content filling a property field. | Q92 (Virginia Woolf) |
| Statement | The complete assertion linking Item → Property → Value. | Q10 → P6 → Q92 |
| Qualifier | An extra detail refining a statement. | start time → 1999 |
9. Check Your Understanding
Test your understanding of structured data and modeling principles with this short, non-persistent self-check:
What does a basic statement in DNKB represent?
Look at the statement: "Orlando: A Biography (Item) ─── author ───> Virginia Woolf (Value)". Which part is the Property?
You are adding the language of a book. The language is French, which already exists as an item (Q51). What value type should you choose?
Which of the following describes a modeling mistake?