The Univeral Knowledge Collaboration Network

The UKCN is a knowledge graph that is similar to other Wikidata-style knowledge graphs. Like existing systems, it: - Hosts data - Contains entities and properties - Offers a range of query interfaces, such as SPARQL and the Knowledge Client - Hosts both well-known objects (e.g., Barack Obama) and obscure ones (e.g., a random uploaded image file) - Largely relies on social processes for data curation

Unlike existing systems, the UKCN also models the computational world, with support for storing files, databases, schemas, execution events, functions, and other items. This allows the UKCN to apply the virtues of traditional knowledge graph systems --- topic flexibility and efficient curation --- to all the data that an application developer might need.

Using the UKCN

Like Wikidata, the UKCN presents an HTML browser interface for users. Anyone can search and retrieve data from the UKCN. With an account, a user can create new UKCN objects, annotate and edit existing ones, run functions, and carry out a range of other tasks.

Every data object in the UKCN has its own URL. This URL can be used in the browser to find an object. The same URL can be provided to the Knowledge Client for various use cases. Sharing data with a colleague is as easy as sharing a webpage: just send your friend the appropriate URL.

Every data object in the UKCN has an affiliated user account as its owner. In the current version of the UKCN, all data objects are world-readable and world-writable. Future versions will implement user permissions that prevent unauthorized access and edits.

Data Model

Data objects in UKCN are immutable: with one important exception, they can never change and are never deleted.

Every immutable data object has an obscure identifier. This can be used to fetch the data object, forever. Because obscure identifers are never recycled, and because data objects can never be modified, UKCN can be used for long-term provenance tracking. An object with an obscure identifier is akin to a particular version of a particular entity in Wikidata; it might be useful and popular, but probably not.

The UKCN also offers well-known identifiers. This is the only kind of value in the UKCN that can change over time. A well-known identifier starts with the letter 'X', followed by an integer. It is akin to the set of top-level entities in Wikidata, such as Q76 (Barack Obama).

A well-known identifier always points to a single immutable data object (which, of course, also has an obscure identifier). This is similar to the way in which a high-level Wikidata identifier (e.g, Q76) at any moment in time points to a particular version of the entity.

In practice, users can mostly ignore this distinction. Most people will mostly deal with well-known objects; that is, objects currently pointed-to by a well-known identifier. Non-well-known objects are generally only of interest when looking at past versions (e.g., for provenance tracking reasons).

Users retrieving UKCN objects using a single query interface, whether its via an obscure identifier or a well-known identifier,

Deployment

There is a single UKCN that is implemented across multiple servers, in the same way there is a single World Wide Web implemented across multiple web servers. This will allow users and institutions to host data that might be tied to a particular location, while allowing users to ignore physical placement when annotating and curating data.

However, the current UKCN software implements only "single-server mode".

Interfaces

The UKCN allows users to access it in three ways: 1. When building applications, via the Knowledge Client 2. When curating and sharing data, via the HTML Browser interface 3. When uploading novel data artifacts programmatically, via the REST API

This last interface is primarily useful when integrating the UKCN with existing data pipelines.

Getting Started

Go here to install and run the UKCN system.