ALEG

This document looks at breaking down some of the large entities described in the data model inventory with more specific descriptions. For example, the work entity covers all types of works, from serials to poems to letters.

The point of this exercise to to resolve issues which get raised when we try to look deeper at what the things we are storing really are - what they consist of and how they interract with other elements. It does not imply a 'physical' implementation strategy, but hopefully progresses us along that path with more confidence.

Work

Fig 1: Work

[MLA: "Main" rather than "Other"?
"Multiwork volume" rather than "Collection" (KF: but a serial Issue is also "multiwork")]

These attributes are common to all works (described in the data model inventory):

• workReference
• title
• variantTitleList
• abstract
• languageList
• ranking
• earliestKnownDate
• worktype MLA: publicationType?
• formList
• genreList
• geographicSettingList
• temporalSettingList
• subjectList
• creatorList
• relatedWorklist
• relatedCreatorList
• otherCreatorsFlag
• manuscriptNotes
• intendedAudience
• usefulForAudience

Serial

A Serial entity contains information about a serial title, eg Westerly.

Serial adds the following attributes to those of work:

• ISSN

  The ISSN of the serial.

Notes and Issues:

• The ISSN of a serial cannot change. A serial can only have one ISSN.

• Serials may be related to other serials via influenced and previouslyKnownAs relationships For example, if serial X changed its name to Y then the two would be linked by a previouslyKnownAs relationship.

• Serials are linked to their individual issues (see SerialIssue below).

• Serials are never instantiated - only their serialIssue's are instantiated

• However, serials may be associated with a URL which points to the 'home page' of the serial

• The serial may have creators associated with it, such as an editor. However, as supported by the general relationship model, editors will come and go and so the editorOf (or will it be ediedBy?) relationship must have a temporal scope (ie, date range) recorded in the relationship (are there other types of scopes? For example (and this is a more general question), could a serial concurrently have many editors specialising in aspects of the publication along geographic or genre "scopes"? If so, is this more effectively dealt with in as a note to the relationship, or by some formal means (as is being proposed with temporal scope) to explicily record the scope in the relationship, mapped to a topic?
  MLA, TW: The latter.

• Is a newspaper considered to be a serial, eg "The Canberra Times"? Do newspapers have ISSN's?
  Yes

• AUSTLIT currently records as a source many variants on "The Canberra Times" and other newspapers such as:
  • The Canberra Times
  • The Canberra Times. 1960
  • The Canberra Times. 1961
  • The Canberra Times (Curtain Up Canberra Playhouse). 1998
  • The Canberra Times (Good Times). 1990
  • The Canberra Times Sunday Times (Fun & Games). 1997
  • The Canberra Times (ANU Reporter). 1996
  • The Canberra Times (Souvenir Edition). 1997
  • The Canberra Times (The Guide). 1998

  Most of these titles represent several issues, but some (such as the Souvenir Edition) don't. How should we handle this information about part of a serial (the TV guide, the children's liftout, etc)? Maybe as:

  • A separate serial, linked to the newspaper by some type of 'partOf' relationship?
  • A separate issue, identified with a title in the serialIssue record?
  • A separate issue, identified with a new serialIssue specific field designed explicitly for this purpose?
  • not at all - the information about where/in-what-section in an issue another work appeared is recorded in the instantiation information for the work - ie, as well as recording the page number, a note is made "in the Good Times section"?

  MLA, TW: The final option is not acceptable.

SerialIssue

A SerialIssue entity contains information about a specific issue of a serial.

SerialIssue adds the following attributes to those of work:

• year
• volume
• number
• month/Season (issue description)

Notes and Issues

• A serialIssue is linked to its "parent" serial via a "issueOf" relationship.

• A serialIssue will usually "inherit" creators from its parent, the serial, by applying those creators active at the time of issue of the serialIssue. However, any serialIssue may have a specific set of creators which apply to that specific issue.

Sequence

A sequence entity just defines a sequence, which is just a name for a group of works related because they are part of the sequence.

Sequence adds the following attributes to those of work:

• sequenceTitle

Notes and Issues

• A Poem Sequence is a special type of sequence, indentified by the workType attribute common to all Works

• A sequence is very much like a collection, except that the works comprising the sequence are very closely related. Maybe the distinction between sequence and collection is not great enough to warrant this separate representation (although it is a convenient place to raise the issues below).

• Is a sequence purely an abstract concept, or could a sequence be instantiated (ie, be published)? If so, would it be published as a collection or as a sequence?
  MLA: Yes and Both.

• A sequence can be ordered or unordered. An ordered sequence needs to record the order of its component works (unlike a collection?): part 1, part 2, etc

• A sequence can be assigned subject classifications not assigned to any of its parts. For example, a sequence on "Summer", "Autumn", "Winter", "Spring" may be assigned the subject "Seasons".

• A component of a sequence could be another sequence (i.e., nested sequences).

• Components of a sequence don't [MLA] necessarily inherit anything from the sequence. I.e., creators, subjects etc are also assigned to a sequence and its components independently.

Collection

A collection entity just defines a work which sets out to combine other works within a single new work. Collections are differentiated as "anthology" and "collected works" and "selected works".

Collection adds the following attributes to those of work:

• collectionType - one of "anthology", "collected works" and "selected works"

Notes and Issues

• A collection would typically be instantiated, perhaps more than once.

• ALEG (and AUSTLIT) records page information for works forming the collection. However, because page information is instantiation related (and not related to the more abstract notion of collection) this information is defined in the relationship record which links to the work to the instantiation of the collection. Somewhat suprisingly, there is no need for a direct relationship between the component work of a collection and the collection.
  
  
  MLA: Perhaps we need to ask ourselves whether we would give page numbers

  Hence, in the above diagram, the link between the verse and the collection (shown as a dotted line) is not actually represented but is derived from the verse appearing in a particular instantiation of the collection. The relationship between the verse and the instantiation of the collection also records the page information.

• Some works in a collection may only appear in the collection and nowhere else.

• Some works may have a special relationship with the collection -eg, a foreword or introduction. These may change substantually between instantiations of a collection. Does this call for another work type: "foreword" and/or "introduction", which gets linked to an instantiation of a collected work with a partOf relationship? I wonder if this relationship is so special/different that the relationship should also annotate this is some way, eg: partOf (foreword). Perhaps relationships will be heavily sub-classed (or specialised) to neatly(?) represent relationship types...

• A collection is independently indexed - it doesn't inherit anything from its components.

Other

The Other sub-class of Work is just the place to classify what is left over - the vast majority of works stored in ALEG.

Other doesn't add any attributes to those of work.

Notes and Issues

• Maybe there is a case to be made for further subdividing this Other into creative/non-creative - I'm not sure. This material is all classified anyway by the form and genre criteria.

Poetry

The poetry entity subclasses Other to hold poetry-specific attributes.

poetry adds the following attributes to those of work:

• firstLineOfPoetry

What about serialised works?

This doesn't really belong here, but somewhere we need to decide how to deal with works (such as a novel) which are serialised. For example, imagine an abridged version of Peter Carey's Oscar and Lucinda was serialized over 3 weekends in The Age.

Firstly, is the abridged version recorded as a new work (who did the abridging - Peter Carey or someone else, and does it matter?) or would abridged versions usually be treated as instantiations annoted appropriately? (I've assumed below it is a separate work, albeit linked to the 'original' by a versionOf relationship.)

Secondly, how best to record exactly how this abridged version has been instantiated? There are at least three possibilities:

1. Create 3 new works, each representing a part of the abridged Oscar and Lucinda. Link each of them to the abridged version with a special "segmentOf" relationship (why not "partOf" - I don't know, maybe, but "partOf" is used elsewhere to group far more unrelated works, such as the individual works which just happen to appear together in a serial Issue). Each of these works is then linked (using "instantiatedAsPartOf") to one of the three instantiations of The Age:
   

2. Link the abridged Oscar and Lucinda (using perhaps "partlyInstantiatedAsPartOf") to one of the three instantiations of The Age. The obvious problem here (although it also occurs in the first approach) is that there is probably a need to link these three partially instantiated links together with yet another relationship so that the fact that it was instantiated in three parts can be easily grasped by the user. Maybe this inter-instantiation relationship could be called "coInstantiation", or something wierd:
   

3. As above, but create a new entity: InstantiationSet specifically to handle this case. The abridged Oscar and Lucinda is instantiated as this one of these InstantiationSets which itself is linked to the three instantiations of The Age:
   

It probably doesn't matter much, but we just have to do something to represent this common case.

KF: For the current thinking on how best to represent this, see the Work Types section of the Relationship Inventory

Creator

Fig 2: Creator

7 June 2000 - what about 'cultural groups' (A.L.M.), eg the Aranda Tribe? another division of Creator, or shoe-horn into 'Organisation'?

The Creator entity is used in ALEG to record information about each creator of works. The above diagram attempts to show that the Creator entity is subclassed by human and organisation.

This isn't a particularly significant dichotomy, but maybe it will help later provide templates or rules for the data maintenance of these entities. For example, the familial relationships of which an organisation could partake are pretty limited!

These attributes are common to all creators (described in the data model inventory):

• creatorReference
• variantNameList
• relatedCreatorList
• biographicalNotesList

Human

A Human entity contains information specific to human creators.

Human adds the following attributes to those of creator:

• name

  broken down into:

  • family name
  • other names
  • title/honorifics (eg, Rev, Sir, Honourable, Ms, ...)
  • display name (if it needs to be displayed in some particular way)

  MLA: and cataloging name

• dateOfApplicability

  Some names may have a date of applicability. For example, and author who changes their name by marriage or deed poll may be best represented with a date range annotating the name. It is proposed that ALEG will represent nee names in this way. Hence, Judith Rodriguez (nee Green) would be represented by two creator records: Judith Rodriguez and Judith Green, perhaps linked somehow by a "alternatively known as" relationship (discussed further below).

  KF: For the current thinking on how best to represent names, see the Names, alternate names, pseudonyms document.

• gender

• birthDetails
  • birthDate
  • birthPlace
  • birthState
  • birthCountry

• deathDetails
  • deathDate
  • deathPlace
  • deathState
  • deathCountry

• nationalityList

• ethnicGroupList

• aboriginal - may include:
  • tribalAffiliationList

• expatriateAustralian (AUSTLIT: aust-e)

• visitorToAustralia (AUSTLIT: aust-v)

• arrivedAustraliaDate

• departedAustraliaDate

• relatedCreatorList

• occupationList

• manuscriptItems/ArchivalItemsList

Notes and Issues:

KF: For the current thinking on how best to represent names, see the Names, alternate names, pseudonyms document.

• It is proposed above that "nee" relationships be represented by linking creator records together with "alternatively known as" relationships. Is "alternatively known as" specific enough, or does this relationship need to record the nature of the alternative name (eg, "nee", "changed by deed poll", "reverted to mother's maiden name"...)

• Pseudonyms will be represented as a "pseduonym of" relationship.

• Does there need to be a "base" name from which "alternatively known as" and "pseduonym of" relationships radiate? Or is a chain (or tree, possibly branching!) a better relationship? If there is a "base" name, does it record some "basic" attributes which are "inherited" by related names, or is each creator record self contained. That is, if creator X is recorded as an expat Australian, does their pseduonym Y and former name Z automatically inherit this expat status, or must it be recorded against one or more versions of their name?

  Imagine this:

  • A person is born as Jane Jones in 1910
  • Between 1927 and 1945 she writes works as herself
  • Between 1940 and 1950 she writes works as "Wild Jane"
  • In 1945, she marries Mr Wills and changes her (writing) name to "Janey Wills"
  • In 1942 she starts to use a pseudonym "Wild Pete" which she continues to use until she dies
  • In 1955, she changes her name by deed poll to "Janet Jackson", but never uses that name to write, except letters to her mother-in-law.
  • Between 1962 and 1965 she uses a pseudonym "Mrs ee cummings"
  • In 1970, she wrote a work as "Jane Jones"

  One way to represent this is as a hub:

  
  Fig 3.

  Another way to represent this is as a chain with cross-links and divisions into multiple chains (because people and names are complex!):

  
  Fig 4.

  Which is "best" (most complete, most useful, most economical, most likely to be accurate)?

Organisation

An Organisation entity contains information specific to non-human creators.

Organisation adds the following attributes to those of creator:

• organisation name

• dateOfApplicability

  (Hmm..this is common to both human and organisation, so should be in the base creator entity...)

Notes and Issues:

• Organisations have previous and subsequent names, can be known as various names at the same time and can have subordinate and superior entities (for example, see how CSIRO is treated in the Australian Science at Work site).

• Can organisations have pseudonyms?
  MLA: 'also known as'

• Is ALEG interested in recording the relationships creators have with organisations? Would ALEG record an organisation which was considered significant, eg, the DLP or Communist Party just so that some of these relationships could be represented (eg, famous author X memberof the Communist Party 1945-1952), or would this be done in some other way (such as making the "Communist Party" a topic (related to the "organisations" topic), and creating a relationship between the "famous author X" and the "Communist Party" topic of type: "memberOf")?
  MLA: Yes (maybe not at first - requires more data than we currently have)