ALEG

Data Model - Relationship Inventory Part 2
Major Topics and Relationships between the them and between major ALEG Entities

To Classify is Human ...

Index

Introduction

This document attempts to enumerate the major topics recorded in ALEG and the relationships between them and the the major ALEG entities (agent, work, expression, manifestation) providing some specific requirements behind the notions and visualisations outlined in the introduction to relationships document.

A companion document describes the relationships between the major entities.

Introduction continued - The Topic Map Approach

The approach being taken, at least for this logical model (which is attempting to describe an abstract, idealised, canonical, "perfect" view of the ALEG data), is to assume that attributes and relationships of data entities can be represented as topics. These topics can be absolutely anything at all - anything that ALEG decides is worth recording.

Topics can be linked together to form associations. For example, the topic "Horsham" could be linked to the topic "Victoria", which in turn could be linked to the topic "Australia". The linking association can be of a type (also a topic!), such as "is part of" in the above example.

Topics each have a "topic type", eg, "town" for "Horsham", "state" for "Victoria", "country" for "Australia". You probably guessed that these 'topic types" are just topics as well!

The name of two topics may be the same, but they may be differentiated by topic type and relationships. For example, the system may record another topic named "Victoria", but that topic may have type of "city" and be associated with ('is part of') a topic "British Columbia" (which in turn is of type 'province').

For more information on Topic Maps, refer to the references listed in the brief introduction to Topic Maps in the ALEG Data Model introduction.

But there are two important points worth emphasing:

  1. Topic Maps are completely extensible. Topics can be added, just as words in a thesaurus can be added.

  2. The semantics (or 'meanings') behind the associations between the topics can be defined to the system in such a way that the system can use them to infer relationships not directly defined. For example, imagine a novel was given a subject of the "Horsham" topic, as defined above. From that assignment and: the system can include that novel in a response to a query "show me all novels set in Victoria".

Going further and allowing more flexability, rather than linking "Horsham" to "Victoria", we could instead:

Now, if the 'fundamental' ALEG entities such as agent and work are linked to "Horsham" (by perhaps "was born in" and "subject of" associations), then the system can answer questions such as "which authors born in the Wimmera region produced works about the Wimmera region", and authors and works will be returned regardless of whether the ascribed place of birth or subject was "Horsham", "Stawell", "Nhill" or, directly, "Wimmera".

As well, if the system knowns that the 'is part of' relationship is transitive (ie, if we know that "A is part of B, and that B is part of C, then we can infer that A must be part of C"), then the system can also find agents born in Victoria and works about Victoria by following the trail from Horsham to the Wimmera to Victoria (well almost - see note just below..).

The following diagram extends the concept even further, by adding the regions "Western Victoria" and "Rural Victoria" (some of the implied 'born in" and 'is part of' relationships which have been omitted for clarity!):

Note:

The 'is part Of' relationship described above is about to be abandoned in this document. "Is part of" is unfortunately ambiguous in everyday speech and can be used to imply a symmetry which does not exist. For example, someone may say "The Great Dividing Range is part of NSW", and even "NSW is part of the Great Dividing Range", and people would understand what they meant. But this representation of the relationship doesn't help us imply anything about an event which might have happened in NSW as to whether it also happened in the Great Dividing Range (we need to know where in NSW it happened), and vica versa.

So, 'is part of' is replaced by 'isContainedWithin' in the more formal specification of relationships below. 'IsContainedWithin' really allows us to apply the transitive operation described above without ambiguity.


A challenge for Topic Map designers is to come up with a set of topics which allows maximum 'bang for the buck': too many topics which people don't use or find confusing is a waste of resources; too few topics and perhaps the required granuality and grouping is not delivered.

For example, in the above Topic Map fragment:

Event Entity

Events store information about things which have happened which can and usually are associated with a date or date range:

Events involve one or more entities/topics.

Event types are themselves 'topics' in the ALEG topic map. And, instances of events are also topics. The instance of the event becomes a topic by inheriting the topic type from the type of the event, and the topic name is contructed automatically based on the topics referenced in the event. For example, the "Birth" event below would have a topic type of "Event, sub-type Birth" and a topic name of "Birth of 'White, Patrick' on 28 May 1912 at Knightsbridge {London, England} ".


Patrick White - The Early Years
A few events sourced from http://www.nobel.se/laureates/literature-1973-1-autobio.html

All Event have a 'Notes' attribute which is not shown below.

  1. birth

    Linked entities/topics:

    Note - a person can have at most one birth event

  2. death

    Linked entities/topics:

    Note - a person can have at most one death event

  3. visited

    Linked entities/topics:

  4. migrated

    Linked entities/topics:

  5. resided

    Linked entities/topics:

  6. influence

    Linked entities/topics:

  7. occupation

    Linked entities/topics:

  8. association

    Linked entities/topics:

  9. creation/realisation/embodiment

    Linked entities/topics:

    Works are "created", Expressions are "realised" and Manifestations are "embodied". Within these creationType topics are subtopics such as 'translated', 'abridged', 'adapted', 'illustrated'.

    Some acts of creation will not have an "input" - for example, the creation of "Voss" the work probably doesn't have any (discernable) input work. Others will - for example, the creation of the work, "Voss the Opera" would have "Voss the novel" as an input, and a creationType of "adapted".

Work Type Topics

This classification is applied to a work to identify works which are in various ways "collections" containing other works, or an extract of another work,

There is only one "work type" topic type: workType.

The topics of type workType are fully enumerated as follows:

Notes:

Form Topics

This classification is applied to a work to describe the "form" of the work. Some people may call this classification "genre", and the ALEG "genre" classification "sub-genre". Oh well.

There is only one "form" topic type: form.

The topics of type form are fully enumerated as follows:

Notes:

Genre Topics

This classification is applied to a work to describe the "genre" of the work. Some people may call this classification "sub-genre". Oh well (again).

There is only one "genre" topic type: genre.

The topics of type genre are fully enumerated as follows:

Notes:

Concept Topics

Concept Topics represent abstract notions or ideas, eg: post-modernism, reconciliation, anger, desire, isolation, war.

Object Topics

Object Topics represent material things, eg: the Mir space station, the Crown Jewels, the Magna Carta.

(What about "The Lodge", "The Dog on the Tucker Box", "The Red Centre" - possible overlap with spatial topics and even concepts???)

Spatial Topics

These topics have a topic type of 'spatial':

  1. transNationalRegion

    Examples:

    Relationships with other spatial topics:

    Relationships Examples:

  2. country

    Examples:

    Relationships with other spatial topics:

    Relationships Examples:

  3. region

    Examples:

    Relationships with other spatial topics:

    Relationships Examples:

  4. state

    Examples:

    Relationships with other spatial topics:

    Relationships Examples:

  5. city

    Examples:

    Relationships with other spatial topics:

    Relationships Examples:

  6. town

    Examples:

    Relationships with other spatial topics:

    Relationships Examples:

  7. suburb

    Examples:

    Relationships with other spatial topics:

    Relationships Examples:

  8. landmark

    Examples:

    Relationships with other spatial topics:

    Relationships Examples:

Issues:

Relationships with other Topics

Other TopicSpatial TopicRelationship Notes
Event all where
Work all subject
Work all setting

Temporal Topics

These topics have a topic type of 'temporal':

  1. date

    Examples:

    Relationships with other temporal topics:

    Relationships Examples:

  2. year

    Examples:

    Relationships with other temporal topics:

    Relationships Examples:

  3. decade

    Examples:

    Relationships with other temporal topics:

    Relationships Examples:

  4. century

    Examples:

    Relationships with other temporal topics:

    Relationships Examples:

  5. era

    An era is a named year range. Bounding years are inclusive.

    Examples:

    Relationships with other temporal topics:

    Relationships Examples:

Relationships with other Topics

Other TopicTemporal TopicRelationship Notes
Event (all) all when
Work (Manuscript Collection) year (range) dateRange documents the date range of material in the manuscript collection
Work all subject
Work all setting

Organisation Topics

Nationality/Identity Topics

Vocation Topics

There is only one "vocation" topic type: vocation

Examples:

Relationships with other vocation topics:

Relationships Examples:

Relationships with other Topics

Other TopicVocation TopicRelationship Notes
Event (occupation) all what


Home > Data Model
Kent Fitch, on behalf of Marie-Louise Ayers, Annette McGuiness and Kerry Kilner
k.fitch@adfa.edu.au
21 June 2000 (pending)
Revised: 4 July 2000