ISO/IEC JTC 1/SC34 N0268

ISO/IEC JTC 1/SC34

Information Technology --

Document Description and Processing Languages

Title: The Ontopia Schema Language: Tutorial
Source: Lars Marius Garshol
Project: TMCL
Project editor: Steve Pepper
Status: Individual contribution
Action: For information
Date: 2001-11-27
Summary: A schema language for topic maps, presented as input for the TMCL process.
Distribution: SC34 and Liaisons
Refer to:
Supercedes:
Reply to: Dr. James David Mason
(ISO/IEC JTC1/SC34 Chairman)
Y-12 National Security Complex
Information Technology Services
Bldg. 9113 M.S. 8208
Oak Ridge, TN 37831-8208 U.S.A.
Telephone: +1 865 574-6973
Facsimile: +1 865 574-1896
E-mailk: mailto:mxm@y12.doe.gov
http://www.y12.doe.gov/sgml/sc34/sc34oldhome.htm

Ms. Sara Hafele, ISO/IEC JTC 1/SC 34 Secretariat
American National Standards Institute
25 West 43rd Street
New York, NY 10036
Tel: +1 212 642-4937
Fax: +1 212 840-2298
E-mail: shafele@ansi.org

Abstract

This document provides a tutorial introduction to The Ontopia Schema Language, suitable for learning the language. For detailed information about the language syntax and semantics, see N267.

Table of contents

1. Introduction

The Ontopia Schema Language provides a way to describe and constrain the information in your topic maps, and this can be made use of in a number of different ways. Some of the most important are:

This tutorial shows you how to write a schema for your topic maps and how to validate the topic maps against the schema, but does not go into making use of the schema in other ways.

2. The basics of the schema language

A topic map schema in the Ontopia Schema Language consists of a set of topic and association class definitions. These class definitions constrain the structure of the instances of the classes, and so control the form information may take in a topic map that uses the schema.

This section shows you how the Ontopia Schema Language works by first giving a general introduction to the concepts behind it, and then walking you through the creation of a simple schema for the 'Free XML Tools' topic map.

2.1. How it works

A schema in the Ontopia Schema Language consists of a set of topic and association class definitions that state what the allowed structure of instances of the classes are. The class definitions contain constraints on the characteristics of instances of the classes, which in turn may contain further constraints.

A topic map is validated against a schema by iterating over the topics in the topic maps, validating each, then repeating the process with the associations. For each topic or association the class it belongs to is found, and the characteristics of the topic or association are then validated against the constraints in the class. Each characteristic is compared with the constraints until a constraint matching the characteristic is found; the characteristic is then validated against that constraint.

Generally, the type of the characteristic is used to match the constraint, but for objects which have no type scope is used. A special case is association roles seen from within topics, which are matched by both role type and association type. This is because different rules may apply to how a topic can play role A in association B from how it may play role A in association C.

The table below shows what objects may be constrained in what contexts, what is used to match them against their constraints, and by what they may be constrained beyond what the rows in the table show.

What can be constrained

Container Containee Element Match by Constrainable
Topic map Topics topic Type Other classes, superclasses
Topic Base names baseName Scope Cardinality
Base name Variant names variantName Scope Cardinality
Topic Occurrences occurrence Type Cardinality, scope, internal/external
Topic Association roles playing Role and association type Cardinality
Topic map Associations association Type Scope
Association Association roles role Type Cardinality, player type

Note that a topic map schema in the Ontopia Schema Language does not contain any topic definitions. Instead, it refers to typing and scoping topics within the topic map to be validated by URI. This can be done by referring directly to the topic element within the topic map to be validated (using topicRef), by referring to the subject indicator of the topic (using subjectIndicatorRef), or by giving the URI of the topic element relative to the document URI of the topic map (using internalTopicRef).

What this means is that schemas in this schema language are not independent of the topic maps they are used to validate. This is slightly awkward, but we have not been able to think of any better solution.

2.2. Introductory example

In this section we'll make a schema for the 'Free XML Tools' topic map, which you can find in the ${basedir}/samples directory. This topic map contains information about free and open source XML tools, their authors, the standards they support, and/or implement, and so on. We'll create the schema step by step, using the validation errors we get to improve the schema.

We start of with the simplest schema imaginable, one that simply says that anything is allowed, which is what the schema below does.

The unconstraining schema


<tm-schema match="loose">
</tm-schema>

This schema has no classes, and uses loose matching. Loose matching means that there are no constraints on topics and associations that are not instances of one of the classes mentioned in the schema. Since this schema has no classes it means that all topics and associations will be accepted.

This, of course, is not very useful, so we'll add a class definition for the 'standard' class of topics. Below is an empty class definition for this topic type.

Empty standard class


  <topic>
    <instanceOf>
      <internalTopicRef href="#TMTT_Standard"/>
    </instanceOf>

    <!-- characteristic constraints go here -->
  </topic>

Above is shown an empty topic class definition. It looks just like a topic in the XTM format, except for the reference to the topic that defines the class. We use the internalTopicRef element here, which contains a URI that is relative to the base address of the topic map, rather than of the schema document. This is useful because it means you can refer to the typing topic without caring where the topic map document is located relative to the schema document.

You can also use the topicRef element from XTM, but this will then depend on where the topic map is located relative to the schema. The subjectIndicatorRef element can also be used, and is independent of where the topic map is located, but requires the typing topic to have a subject indicator.

If we try validating the topic map with this schema (see section 3.1) we get 374 errors, so clearly we need to improve the schema. One thing we do know is that all standards have at least one base name in the unconstrained scope. We add the following constraint to reflect that, causing the topic class definition to look as shown below.

Standard class with basename


  <topic>
    <instanceOf>
      <internalTopicRef href="#TMTT_Standard"/>
    </instanceOf>

    <baseName min="1" max="Inf">
      <scope></scope>
    </baseName>
  </topic>

This example says that the topic must have one base name or more in the unconstrained scope (the scope element is empty). The min attribute says we must have at a minimum 1 base name matching this constraint, while the max says that there is no upper limit on the number of base names. Note that if we'd left out these two attributes there would still have been no upper limit, but standards would then be allowed to have no name in the unconstrained scope.

Validating with this constraint we get only 290 errors. Many of these are related to occurrences of the types 'specification', 'home page', 'link collection', and 'informational site'. We add four new constraints to the topic class in order to get rid of those errors.

Constraints for occurrences


    <occurrence external="yes">
      <instanceOf>
  	<internalTopicRef href="#TMOR_Specification"/>
      </instanceOf>
    </occurrence>

    <occurrence external="yes">
      <instanceOf>
  	<internalTopicRef href="#TMOR_Homepage"/>
      </instanceOf>
    </occurrence>

    <occurrence external="yes">
      <instanceOf>
  	<internalTopicRef href="#TMOR_Site"/>
      </instanceOf>
    </occurrence>

    <occurrence external="yes">
      <instanceOf>
  	<internalTopicRef href="#TMOR_Links"/>
      </instanceOf>
    </occurrence>

These constraints allow topics of class 'standard' to have occurrences of these four types. The external attribute is here used to say that the occurrences must be external (that is, they can't use resourceData). It is possible to constrain the scope of occurrences as well, by using the scope element inside the occurrence element, but we have no use for that here, so we don't do it.

Validating with this improved schema gives us 247 errors, all of them related to standards playing the role 'standard' in associations of the type 'product implements standard' or 'product uses standard'. We solve this by adding a constraint allowing these two uses.

Constraint for association roles


    <playing>
      <instanceOf>
  	<internalTopicRef href="#TMTT_Standard"/>
      </instanceOf>
      <in>
        <instanceOf>
    	  <internalTopicRef href="#TMAT_StandardImplemented"/>
        </instanceOf>
        <instanceOf>
    	  <internalTopicRef href="#TMAT_ProductUse"/>
        </instanceOf>
      </in>
    </playing>

This element says that topics of type 'standard' may play this role in associations of one of the given types.

When we add this constraint to the schema it validates with no errors at all. Note, however, that so far we only validate topics of one type, and we do not validate associations at all. We validate the roles the topics of type 'standard' play, but not the associations themselves. In order to allow associations to be constrained as well the schema language also supports association class definitions

As an example, we'll create an association class definition for the 'standard implemented' association. This is given below.

Association class definition


  <association>
    <instanceOf>
      <internalTopicRef href="#TMAT_StandardImplemented"/>
    </instanceOf>

    <role min="1" max="1">
      <instanceOf>
  	<internalTopicRef href="#TMTT_Product"/>
      </instanceOf>
      <player>
  	<internalTopicRef href="#TMTT_Product"/>
      </player>
    </role>

    <role min="1" max="1">
      <instanceOf>
  	<internalTopicRef href="#TMTT_Standard"/>
      </instanceOf>
      <player>
  	<internalTopicRef href="#TMTT_Standard"/>
      </player>
    </role>
  </association>

What this class definition does is to say that associations of this type must contain one role of type 'product', the player of which role must be of the same type, and one role of type 'standard', and that the player of this role must also be of type 'standard'. This means that the association is constrained to always be a simple binary association between topics of specified types.

At this stage, of course, we only validate one class of topics and one class of associations. If we now change the schema to use strict matching we get 1154 errors when validating.