Published Subjects Terminology
Proposed by Lars Marius Garshol
January 26, 2002

This is a working document for the PubSubj TC, which explains the terminology relating to published subjects from one particular point of view. There are other points of view within the TC, so this document is in no way official.
(Thanks to Sylvia Schwab for a very helpful outside view.)

SUBJECTS AND THEIR PUBLISHERS

While reading it may be useful to keep an eye on this diagram: http://www.ontopia.net/tmp/terms.png

Imagine that someone wants others to be able to refer to a certain concept in an unambiguous fashion in a global context. In topic map terms, concepts are called /subjects/, but they have no obvious identifiers, in topic maps or out of them. So, what to do? One solution is to publish a /subject definition resource/ (SDR) for the subject. This is an information resource[1] that defines clearly (to a human reader) what the subject is. The publication of the SDR makes the subject a /published subject/, and the interested party the /publisher/ of that published subject. Note that the published subject is the real-world concept we want to make statements about, not what the publisher publishes. That is the SDR, and its role is to explain what the published subject is. Those who want to refer to the subject can now do so by referring to the SDR.

PUBLISHED SUBJECTS IN TOPIC MAPS

Now, someone wants to make a topic map where they make statements about the published subject. To do this the topic map author needs to make a topic in the topic map and give that topic a /subject indicator reference/ (SIR), which points to the SDR. Doing so makes the SDR a /subject indicator/ in this topic map. Note that any resource can be used as a subject indicator. All that's necessary for this to happen is for someone to refer to it as a subject indicator from some topic map. This is somewhat risky, however, as the resource may not be completely unambiguous as to what subject it defines, but for want of a formal SDR one may well use some resource that seems reasonable. Part of the reason to use published subjects is so that topic maps will merge correctly even when written by parties unaware of one another. When merging topic maps the topic map software will not download and compare the subject indicators, however, and therefore it's vital that all references to SDRs use exactly the same SIRs. For this reason the publisher must not only publish the SDR, but also clearly define the URI to be used to refer to the SDR.

DOCUMENTING PUBLISHED SUBJECTS

In practical terms, what this means is that the publisher needs to publish more than just a set of SDRs. (We spoke initially of a single subject, but subjects will rarely, if ever, be documented singly.) The complete package of documentation published is called the /published subjects documentation/ (PSD). This will generally contain metadata about the publisher, the collection of published subjects and references to each SDR. In addition, it should define the canonical SIR to be used for each published subject, and perhaps also some assertions about each published subject (such as names and topic types). There are no definite rules for what PSDs must contain, though this TC is working to define recommendations for their contents.

SUMMARY

A published subject is a subject for which someone has published an SDR. An SDR is an information resource published in order to clearly define a subject for a human reader. Subjects which have no SDRs are not published subjects. Publishers are recommended to establish sets of published subjects by publishing a PSD package, in order to establish the set of subjects, metadata about the set and the SIR of each subject. From the point of view of a topic map any information resource used to indicate the identity of a subject is a subject indicator. The subject indicator may be a SDR, if the subject is a published subject, or it may not be. The URI used to refer to the subject indicator is known as the SIR.

[1] This is what is known in RFC 2396 as a network-retrievable resource.