ALEG

What I've done

• Attempted a load of the LAW data file by matching surname, firstname, date-of-birth, date-of death against AUSTLIT data (ie, a fairly "exact match"). Then, matched as above, but just with first initial. Then matched surnname,firstname and either date of birth or date of death. Then matched surname,firstname with a date of birth or death plus of minus 3 years. Then matched surname, firstname. In all cases, if more than one match was found, the record was not matched (as it was ambiguous).

  Of the approximately 10,160 LAW entries, about 850 ambigutities were identified and about 3890 records were "positively" matched. This left and 6270 records to be processed. Ben has been working on these during the week and will be for the next few weeks!

• The maintenance of works/expressions/manifestations is now operational. although rough around the edges and incomplete. However, the main editing functions and updating of the database now works (at least for my test cases!). There is much to do, including:
  1. new attributes, including rank, periodical (for periodical issue)
  2. work-specific edits. For example, the system should enforce that periodical details and parent periodical work are both entered if and only if the work is a periodical issue.

  I'm planning to implement the work-specific edits by applying XSLT rules to the document.

  As an aside, XSLT (XML Stylesheet Language - Transformation) was intended to be a flexible language for transforming one XML document into another, and indeed it is. As with many good tools, XSLT has been used in innovative ways which were not forseen by the XSLT designers. One of these is for document verification - ie, the job of applying edit criteria to a document to check whether it conforms to some abstract model. The W3C, managers of the web/XML standardisation processes, have produced a standard for defining the definition of any particular XML document. This standard "XML Schema" has a long and controversial history. People want to be able to define lots of things which make a document "valid" on their context, and demand a lot from an XML Schema standard, which in turn makes it complicated, hard to understand and implement and maybe even hard to use.

  Well, Rick Jelliffe took a completely different approach to testing a document for validity. Rather than compare it with a declaration (the XML Schema approach), his Schematron uses the XSLT language to define a set of rules which are tested against an incoming document. Basically, it transforms the incoming document by using the rules to produce a report which describes whether or how well the incoming document matches the specification defined by the rules. The Schematron is a simple and brilliant idea.

  Some of the nice things about using XSLT to define the "validity" rules are:

  1. The rules aren't limited by the syntax of a declarative schema language. For example, if you really need to test that if the 3rd "food" element contains the string "Lemon" then the "heading" element must contain the string "Citric" or "Citrus", then you can easily do it with the Schematron approach.
  2. The edit rules are removed from code into a file which knowledgable users can read and understand. XSLT isn't the simplest language in the world, but you don't have to be a programmer to understand the basics and then explore the more esoteric.
  3. XSLT is a extremely important cornerstone of the XML world. It is very likely to be widely implemented, supported and understood for a long time.

• Started working on the next iteration of the search and display user interface. There was a flurry of emails on the specifications for this a few weeks ago, and I hashed out some concrete details with Marie-Louise and Annette on Friday (12th), and started implementing it.

  Unfortunately, we had a large power outage at ADFA just after 5pm Friday (which is why this weekly summary is late), and the Uninterruptable Power Supply (UPS) which powers the SUN box failed (don't ask me why they are called "Uninterruptable" - over the years I've noticed that UPS's are the single major cause of outages...).

  So, I lost a few hours work, but the system was brought back up early Monday morning by Fran (who bypassed the UPS - good on ya Fran! Maybe it will be replaced with a power-surge protector from K-Mart!)

• Add more attributes to the work/expression/manifestation edit schema.
• Add general edits based on XSLT Schema (the Schematron approach).
• Add author/title as subject.
• Add topic selection to the work/expression/manifestation UI

• Implement the next iteration of the search and display user interface

• Keep improving the maintenance user interface

• Our new server was scheduled for delivery in late December. The latest update from Sun Microsystems asserted that it would arrive soon, possibly on Jan 15th. Setup and transfer of the system onto this server will probably be 2 days work.

• Things are going well. I just wish I'd remember to click the "save all" button a few times on Friday arvo....