| TITLE: | Draft Alignment of N0396 with N0393 |
| SOURCE: | Jan Algermissen |
| PROJECT: | Topic Maps |
| PROJECT EDITORS: | Michel Biezunski, Martin Bryan, Steven R. Newcomb |
| STATUS: | Editor's Draft, Revision 1.3 |
| ACTION: | For review and comment |
| DATE: | 27 April 2003 |
| SUMMARY: | |
| 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-mail: mailto:[email protected] http://www.y12.doe.gov/sgml/sc34/sc34oldhome.htm Ms. Sara Hafele Desautels, 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: [email protected] |
27 April 2003
| 0 | Introduction |
This document is an effort to align N0396 (the current draft of the "Standard Application Model") with N0393 (the current draft of the "Reference Model"). For such alignment it is neccessary to separate the structural material of N0396 from the material that defines semantics (e.g., 'Occurrence-ness') over and above those defined in N0393. Once this semantic material is identified, it can be expressed as a Topic Map Application (TMA) definition as described in section 6 of N0393. Annex A provides a draft of such a TMA definition using an illustrative XML syntax.
The alignment of the variant name structure is omitted, because it follows logically from the alignment of the other structures.
I hope that this document will contribute to a better understanding of N0396 and N0393 and that it will help to resolve open issues in both documents.
| 1 | Comparision of N0396 and N0393 |
| 1.1 | Similarities |
N0396 is similar to N0393 in that it uses surrogates for subjects and typed connections between these surrogates to provide a representation of information.
It is similar in the way it uses properties of the items to identify what the subject is that an item surrogates.
| 1.2 | Differences |
The data model of N0396 uses typed surrogates for subjects, which prohibits the merging of surrogates (of different types) even if they are known to surrogate the same subject. N0396 introduces the [reifier] and [reified] properties to solve this situation.
Typed surrogates add unnecessary complexity to the underlying data model, although the concept of names, occurrences and associations as types might be useful when designing an API on top of N0393.
The model defined in N0393 does not use typed surrogates, thus enabling subjects to be always represented by a single surrogate. This approach is less complex, while it does not constrain the design of APIs for interacting with a topic map.
| 2 | Overview of the alignment |
In order to align N0396 with N0393 the following issues need to be resolved:
What properties in N0396 are intended to define the subject identity of information items? These properties can be directly expressed as SIDPs in the TMA definition.
How can the different kinds of relationships between subjects (names, occurrences, and so on) be expressed using the assertion structure of N0393? What assertion types and role types need to be defined in the TMA definition?
How can the properties in N0396 that are not used for subject indication be defined as OPs in the TMA definition?
| 3 | What properties in N0396 are intended to define the subject identity of information items? |
The properties [subject addresses], [subject identifiers] and [source locators] are properties that define the subject that a given surrogate represents. They map directly to SIDP definitions in the TMA.
| 4 | How can the different kinds of relationships between subjects (names, occurrences, and so on) be expressed using the assertion structure of N0393? |
The data model defined in N0396 does not enable the representation of naming and occurrence characteristics as assertions, since the subjects that play the name role or the occurrence role are not recognized as such and are therefore not surrogated. In the case of names, the name is just represented as a string valued property; in the case of occurrences, the information resource is just represented as a locator valued property.
However, the prose of N0396 says that both, name and occurrence
characteristics are indeed associations:
"Occurrences are essentially a specialized kind of association, where
one participant in the association must be an information resource."
(Section 3.7)
"Essentially, a base name is a specialized kind of occurrence."
(Section 3.5)
Below I propose a small change to the model defined by N0396 that enables the integration of two item types 'occurrence item' and 'name item' with the item type 'association item'. This change does not only makes the alignment of N0396 with N0393 more efficient, it also reduces the complexity of the model of N0396 itself.
| 4.1 | Proposed change to N0396: |
Change the value of the [name] property on the name item to be a topic (a topic that surrogates the information resource that is the name string) and change the value of the [resource] property on the occurrence item to hold the topic that surrogates the information resource that plays the role of occurrence. Introduce a new property on the topic item, called [subject data] that can hold the data content of an information resource. Finally, drop the [value] property from the occurrence item since that value can now be 'stored' in the [subject data] property of the topic that represents the information resource.
This change provides the additional benefit that each information resource addressed from within the topic map will be represented as a topic and thus all information about that resource will be readily accessable from this single location, again in aid of the Subject Location Uniqueness Objective.
With this change it is possible to integrate the different types of topic characteristic with the concept of association, thus significantly reducing the complexity of the model.
By simply providing the appropriate assertion types and role types, all relationships between subjects defined in N0396 can be expressed with the assertion structure defined by N0393.
| 4.2 | What assertion types and role types need to be defined in the TMA definition? |
For the topic-name, topic-occurrence, and class-instance relationships one assertion type and two role types each have to be defined in the TMA definition. The semantics of these relationships include that there is only one player playing each role.
Assertion types and role types for the supertype-subtype relationship must also be included.
| 4.3 | How can the properties in N0396 that are not used for subject indication be defined as OPs in the TMA definition? |
N0396 defines one property ([scope]) that needs to be expressed as an OP in the TMA definition. As in N0396, the TMA-defined property has as its value the set of topics that are the scope.
| 4.4 | Remaining issues |
The [type] property could also be expressed as an OP (having the typing topic as its value) but since the notion of class-instance-ness will be eventually part of the SAM semantics anyway, I chose to map the [type] property to an assertion of type class-instance.
The remaining properties of N0396 are entirely structural properties, and hence redundant given N0393. Some of them map directly to the 'TMM defined properties' of N0393 (e.g. [roles played] ) while the others become obsolete (e.g. [reifier] and [reified]).
| Annex A | Annex A |
This Annex provides a TMA definition for the semantic material of N0396 using an XML syntax devised to illustrate such a definition.
<?xml version="1.0"?>
<!DOCTYPE tma
[
<!ENTITY base "http://www.isotopicmaps.org/psi/">
]>
<tma>
<name>IS13250-N0396-1.0</name>
<description>
A TMA definition of N0396.
</description>
<requiredModels />
<valueTypes>
<valueType>
<name>topic</name>
<description>
A value of the type topic is a surrogate for a subject. The exact
nature of a topic is entirely in the realm of the application. It
could be an integer value providing a unique ID, the address of a
data structure, or a data structure, etc.
</description>
<equality>
Two values are equal if the application considers them to be equal.
</equality>
<merging>
Values may not be merged.
</merging>
</valueType>
<valueType>
<name>locator</name>
<description>
A value of the type locator consists of a pair of non-empty strings, one
containing the name of the notation, the other containing the
reference.
</description>
<equality>
Values are equal if the first strings of the pairs are byte identical and
the second strings of the pairs are byte identical.
</equality>
<merging>
Vales may not be merged.
</merging>
</valueType>
<valueType>
<name>text</name>
<description>
A value of type text is an arbitrary sequence of bytes.
</description>
<equality>
Two texts are equal if they are represented by the same sequence of
bytes.
</equality>
<merging>
Vales may not be merged.
</merging>
</valueType>
<valueType>
<name>set</name>
<description>
A set is a unordered collection of values. All values must have the same
value type and no two values may be equal.
</description>
<equality>
Two sets are equal if the contain exactly the same elements.
</equality>
<merging>
The result of merging two set values is the union of the two sets.
</merging>
</valueType>
</valueTypes>
<properties>
<property>
<name>PR_SubjectAddresses</name>
<valueType>set(locator)</name>
<type>SIDP</type>
<description>
A set of locator items. The locators, if present, refer to the
information resource that is the subject of this topic. If the
set contains more than one locator this implies that the locators
all address the same information resource.
The fact that in some information management systems (e.g., the Web)
an information resource is not directly addressable is ignored.
</description>
<mergingCondition>
Two topics exhibit a value for this property and the
intersection of the two sets is not empty.
</mergingCondition>
</property>
<property>
<name>PR_SubjectIdentifiers</name>
<valueType>set(locator)</name>
<type>SIDP</type>
<description>
A set of locator items. The locator items refer to the subject indicators
of this topic.
</description>
<mergingCondition>
Two topics exhibit a value for this property and the
intersection of the two sets is not empty.
</mergingCondition>
</property>
<property>
<name>PR_SourceLocators</name>
<valueType>set(locator)</name>
<type>SIDP</type>
<description>
If the existence of a topic has been demanded by the presence of
some syntactical construct (a node demander), the address of that
syntactical construct is said to be a source locator of that topic.
Merging can cause the situation that a single topic has multiple
source locators, thus this property is a set.
</description>
<mergingCondition>
Two topics exhibit a value for this property and the
intersection of the two sets is not empty.
</mergingCondition>
</property>
<property>
<name>PR_SubjectData</name>
<valueType>text</name>
<type>OP</type>
<description>
A topic that exhibits a value for this property is the surrogate
for an information resource. The data content of the surrogated
information resource is the value of this property.
This is an OP, because the data content of an information resource
is not used for subject discrimination. Two information resources
can have the same data content and still be different information
resources.
</description>
</property>
<property>
<name>PR_Scope</name>
<valueType>set(topic)</name>
<type>OP</type>
<description>
Topics that exhibit a value for this property must be surrogates for
relationships (that is, they must exhibit a value for the property
IS13250::a-sidp). The set of topics that is the value is defined
to represent the context in which the represented relationship is
considered to be valid.
</description>
</property>
</properties>
<assertionTypes>
<assertionType>
<name>AT_NamedName</name>
<subjectIdentity builtInTopic="#t1" />
<roleType>
<name>RO_Named</name>
<subjectIdentity builtInTopic="#t2" />
<rolePlayerConstraint>
There may only be one player of this role in a single assertion.
</rolePlayerConstraint>
</roleType>
<roleType>
<name>RO_Name</name>
<subjectIdentity builtInTopic="#t3" />
<rolePlayerConstraint>
There may only be one player of this role in a single assertion.
Players of this role must exhibit a value for at least on of the
two properties PR_SubjectAddress and PR_SubjectData, that means,
they must be surrogates for information resources.
</rolePlayerConstraint>
</roleType>
</assertionType>
<assertionType>
<name>AT_OccurringOccurrence</name>
<subjectIdentity builtInTopic="#t4" />
<roleType>
<name>RO_Occurring</name>
<subjectIdentity builtInTopic="#t5" />
<rolePlayerConstraint>
There may only be one player of this role in a single assertion.
</rolePlayerConstraint>
</roleType>
<roleType>
<name>RO_Occurrence</name>
<subjectIdentity builtInTopic="#t6" />
<rolePlayerConstraint>
There may only be one player of this role in a single assertion.
Players of this role must exhibit a value for at least on of the
two properties PR_SubjectAddress and PR_SubjectData, that means,
they must be surrogates for information resources.
</rolePlayerConstraint>
</roleType>
</assertionType>
<assertionType>
<name>AT_ClassInstance</name>
<subjectIdentity builtInTopic="#t7" />
<roleType>
<name>RO_Class</name>
<subjectIdentity builtInTopic="#t8" />
<rolePlayerConstraint>
There may only be one player of this role in a single assertion.
</rolePlayerConstraint>
</roleType>
<roleType>
<name>RO_Instance</name>
<subjectIdentity builtInTopic="#t9" />
<rolePlayerConstraint>
There may only be one player of this role in a single assertion.
</rolePlayerConstraint>
</roleType>
</assertionType>
<assertionType>
<name>AT_SuperclassSubclass</name>
<subjectIdentity builtInTopic="#t10" />
<roleType>
<name>RO_Superclass</name>
<subjectIdentity builtInTopic="#t11" />
<rolePlayerConstraint>
There may only be one player of this role in a single assertion.
</rolePlayerConstraint>
</roleType>
<roleType>
<name>RO_Subclass</name>
<subjectIdentity builtInTopic="#t12" />
<rolePlayerConstraint>
There may only be one player of this role in a single assertion.
</rolePlayerConstraint>
</roleType>
</assertionType>
<assertionTypes>
<builtInTopics>
<topic id="t1">
<property model="IS13250-N0396-1.0" name="PR_SubjectIdentifiers">
<set>
<locator notation="URI">&base;at-named-name</locator>
</set>
</proerty>
</topic>
<topic id="t2">
<property model="IS13250-N0396-1.0" name="PR_SubjectIdentifiers">
<set>
<locator notation="URI">&base;role-named</locator>
</set>
</proerty>
</topic>
<topic id=t3">
<property model="IS13250-N0396-1.0" name="PR_SubjectIdentifiers">
<set>
<locator notation="URI">&base;role-name</locator>
</set>
</proerty>
</topic>
<topic id=t4">
<property model="IS13250-N0396-1.0" name="PR_SubjectIdentifiers">
<set>
<locator notation="URI">&base;at-occurring-occurrence</locator>
</set>
</proerty>
</topic>
<topic id=t5">
<property model="IS13250-N0396-1.0" name="PR_SubjectIdentifiers">
<set>
<locator notation="URI">&base;role-occurring</locator>
</set>
</proerty>
</topic>
<topic id=t6">
<property model="IS13250-N0396-1.0" name="PR_SubjectIdentifiers">
<set>
<locator notation="URI">&base;role-occurrence</locator>
</set>
</proerty>
</topic>
<topic id=t7">
<property model="IS13250-N0396-1.0" name="PR_SubjectIdentifiers">
<set>
<locator notation="URI">&base;at-class-instance</locator>
</set>
</proerty>
</topic>
<topic id=t8">
<property model="IS13250-N0396-1.0" name="PR_SubjectIdentifiers">
<set>
<locator notation="URI">&base;role-class</locator>
</set>
</proerty>
</topic>
<topic id=t9">
<property model="IS13250-N0396-1.0" name="PR_SubjectIdentifiers">
<set>
<locator notation="URI">&base;role-instance</locator>
</set>
</proerty>
</topic>
<topic id=t10">
<property model="IS13250-N0396-1.0" name="PR_SubjectIdentifiers">
<set>
<locator notation="URI">&base;at-superclass-subclass</locator>
</set>
</proerty>
</topic>
<topic id=t11">
<property model="IS13250-N0396-1.0" name="PR_SubjectIdentifiers">
<set>
<locator notation="URI">&base;role-superclass</locator>
</set>
</proerty>
</topic>
<topic id=t12">
<property model="IS13250-N0396-1.0" name="PR_SubjectIdentifiers">
<set>
<locator notation="URI">&base;role-subclass</locator>
</set>
</proerty>
</topic>
</builtInTopics>
<propertyValueComputationRules>
<!--
This TMA does not define any rules for conferring values on
properties on the basis of the roles that a topic plays.
-->
</propertyValueComputationRules>
</tma>
|