AustLit

What I've done

• More thinking on adding scope information to relationships. As discussed last week, adding scope to relationships is a significant enhancement because it allows us to qualify relationships between topics, where relationships are typically "assertions" with an RDF-style predicate, subject and object such as:
  • "gender", "some topic representing an agent", "some topic representing a gender", or
  • "hasCommonName", "some topic representing an agent", "some topic representing some text which can be interpreted as 'surname, other names'"

  Side note: the roles played by the subject and object with the predicate cause me confusion - I have to guess hard to work out whether the 3-tuple: {"usesPseudonym", "agent A", "agent B"} means "A is a pseudonym of B" or "B is a pseudonym of A". Hence, I gravitate to the explicit, long-winded topic-map style with nice, clear roles:

  <topic id="agentA" /> <topic id="agentB" /> <topic id="pseudonymRelationship" /> <topic id="usesPseudonym" /> <topic id="usedAsAPseudonym" /> ... <association> <instanceOf> <topicRef href="#pseudonymRelationship"/> </instanceOf> <member> <roleSpec> <topicRef href="#usesPseudonym"/> </roleSpec> <topicRef href="#agentA"/> </member> <member> <roleSpec> <topicRef href="#usedAsAPseudonym"/> </roleSpec> <topicRef href="#agentB"/> </member> </association>

  The only explicit qualification we currently support is an "uncertain" flag. Qualification can be done by adding notes, but it gets a bit messy.

  Initially we want to qualify subject-as-award with a date, but there are other obvious candidates such as qualifying use of a alt-name or a pseudonym over a date range, or explicitly adding a note or a source to an relationship asserting a cultural heritage.

  As mentioned, adding this capability to the database and xml generation routines is very simple. But I've been experimenting (OK, lets say procrastinating) with various ways to add it to the xml representation and none are ideal.

  Last week I was favouring adding "scope" as a child element to the relationship element, and was hand-wringing about the mixed content which would ensue with the text of the relationship (eg "Female") being "mixed" as a sibling of the "scope" element. The more I played with trying to adapt the user maintenance function to handle this, the less attractive it looked - that code is already complicated enough!

  Another way is to just reference the scope in the relationship:

  <topic ... > <attributes> <gender seq="0" tid="z#" topicType="z" typeCode="P1" uncertain="n" scopeId="s1"> Female </gender> ... </attributes> <scopes> <scope tid="s1"> <hasDate seq="0" tid="VQX" topicType="V" typeCode="V" uncertain="n"> <date startYear="1962" endYear="1992" tid="VQX">1962-1992</date> </hasDate> <hasNote seq="1" tid="Tx" topicType = "T" uncertain="n"> Had sex change operation whilst visiting Peru in 1962, reversed in London in 1992. </hasNote> </scope> ... </scopes> ... </topic>

  This is neat in that it has a minimal impact on the code that processes the relationships, and it is only slightly harder to process in stylesheets. But it does seem a bit of a hack - I think ideally scopes would be represented in the XML immediately under the element they are scoping.

  Hopefully I'll bite the bullet and implement scope next week.

• Added functionality to specify a work award year in the advanced search. So now, you can search for works awarded (or joint winner, or runner up, or ...) the Miles Franklin between 1965 and 1970, or all works awarded an award with a 1933 'date', etc.

• Discussions with Kerry and Annette on what should be shown in the public view. Following these, I've tentatively implemented a public author browse interface which allows about 3% of the agents in AustLit to be listed alphabetically selected by first letter of surname. Summaries and shown, and members of the public can view the agent details, but not works. Agents featured on the home page (with birth/death anniversaries) can also be viewed now by non-subscribers.

  The intent is to get author information on authors of common interest indexed on Google (etc) so that AustLit becomes well known as an authority on Australian Literature. The authors in the browsable list all have birth and/or death events, a bio and 3 or more works (although these criteria are currently being discussed). It is a balancing act between attracting subscribers to pay for the service and becoming so hidden behind the subscription barrier that potential subscribers don't find out about AustLit.

• Misc and many minor changes to stylesheets and other formatting following maintainer and user feedback (use override date for periodical issue year, treat "related to work" symmetrically (wasn't showing issue of 'object' work when that work was a periodical issue), correctly formatting thesaurus scope notes which included markup, show subjective ranking for source works on work display summary and detail representations, display ISSN, show all alt-titles at the work level, show "Non-Austlit" indicator in maintenance agent search results, ignore commas in agent name in maintenance agent search, moved work note to display before expression/manifestations in work detail stylesheet, support "publisher" as an agent role in the creation event, changed ejournal/periodical/newspaper templates to show creation-publisher/place/date by default)

• The system was stable.

Next Week

• Continue design/implementation of relationship scopes. The first use for this facility will be year-scoping of award-as-subject relationships.

• Move awards to be event-based rather than work-based.

• Finalise "public view" processing.

• Multiple creation events for a work as a mechanism for allowing date ranges to be associated with agents responsible for works, eg editors of a periodical.

• Refining NBD Holdings searches.

• Review all subset definitions for efficacy.

• Import/export in MARC and DC formats.

• SDI facility.

• Combining searches

• System Documentation