Skip to content
This repository has been archived by the owner on Nov 20, 2023. It is now read-only.

URIs

Jump to bottom
Matt Wallis edited this page Apr 21, 2016 · 3 revisions

URIs

To publish linked open data, we must mint URIs. This page describes the way this is done for the data published by the Institute for Solidarity Economics. We will be following the suggestions for Creating URIs made by data.gov.uk.

Update, April 2016

We've received advice not to follow the 303 redirection method to separate real-world (non-web) things from documents (web) things, and the grounds that it makes things more complicated than necessary, and that it is fragile (e.g. URIs copied from browser locations are always the URI of the thing-on-the-web, and can easily be pasted into places which should be using the URI of the thing-not-on-the-web).

The article what do uris mean anyway by Jeni Tennison is interesting.

Domain

ISE will publish URIs in the domain data.solidarityeconomics.org.

Separate URIs for real-world things and for documents

URI Description
http://data.solidarityeconomics.org/id/x/y/z/1536 URI of real world thing
http://data.solidarityeconomics.org/doc/x/y/z/1536 URI of document to represent it

In both cases, content-negotiation will be used to respond to the request for one of the above URIs with a document that may be named, for example x/y/z/1535.html or x/y/z/1536.rdf.

x/y/z/ does not mean that there are always 3 levels of classification. Usually, there will be fewer. Thes levels of classification are related to the types of object for which we will be publishing data.

Types of published object

We consider following the data.gov.uk pattern for URIs (e.g. http://education.data.gov.uk/id/school/520965) in which a term like school is used. To do so, we need to identify disjoint sets (i.e. no element in common) to substitute for the term school used above.

At the time of writing, there are two types of object for which we want to publish data:

  1. Initiatives within the Social and Solidarity Economy.
  2. Co-ops extracted from the Co-ops UK data set.

Clearly, there is an overlap between these sets (i.e. they are not disjoint), but neither is a subset of the other: there are SSE initiatives that are not co-ops, and we cannot expect that all co-ops would be treated as initiatives within the SSE.

In the longer term (and in an ideal world!), we would not want to be publishing data about co-ops that cannot be regarded as part of the SSE. However, there is no test available to determine if an initiative is part of the SSE.

Aside (not strictly about URIs): This begs the question of how the DCAP-SSE class SSEInitiative should be interpreted, given the lack of a test for 'membership' of the SSE. This subject needs further investigation.

As a temporary measure, in order not to block the development process, we will use the term experimental in our URIs: http://data.solidarityeconomics.org/id/experimental, as a top-level classification.

The classification experimental indicates that one or more of the following conditions may apply to the data:

  • the use of the DCAP-SSE class SSEInitiative does not imply that the initiative has passed any test of 'membership' of the SSE.
  • persistence of URIs cannot be guaranteed. This limitation should be eliminated as soon as possible.
  • duplicated URIs may exist for the same real-world initiative under the experimental URI classification.

URIs for Co-ops UK objects

We will be using the Co-ops UK dataset to generate examples of Linked Open Data. We need to make it clear that the objects referred to by these URIs are Co-ops registered with Co-ops UK. In particular, we are not expressing any view that these objects (co-ops) are part of (i.e. satisfy the principles and values of) the Solidarity Economy. Rather, we are using them as an experimental data set. So, we will use the following form of URI:

URI Description
http://data.solidarityeconomics.org/id/experimental/co-ops-uk/1536 URI of real world thing
http://data.solidarityeconomics.org/doc/experimental/co-ops-uk/1536 URI of document to represent it

URI persistence

See Cool URIs don't change by Tim Berners-Lee. See related issue.

Discussion

Consider the following scenario: We take the Co-ops UK open data set, and create linked data representations of some of the Co-ops listed therein, because we want to publish data about them as SSE initiatives. We will use a program to convert from .csv. to .rdf, to extract the information we want. In the course of doing so, we must associate each co-op with one of our URIs. Now suppose we perform the same conversion task next year, to bring our data up to date. We must ensure that the URI that we have assigned to a particular co-op does not change.

Now suppose that a further year later Co-ops UK starts publishing its dataset as linked open data, and that there is no longer a need for us to convert it. We must still maintain the persistence of our URIs. If we do not do this, then we will break links that other people have made to our data.

Clone this wiki locally