4. Database history

In 2008, Tarrin Wills began redesigning all aspects of the database to fit the relational data model. This involved restructuring the encoding of stanzas, prose order information, translations, kennings and other materials into a series of interconnected database tables. It also involved updating the web interface to facilitate data entry and editing of this material.

The new system was tested in 2008-9, when volume II of the project, using the new text encoding structure, was successfully exported for print publication.

In 2009, the various font encodings used by early forms of the database (the ReykholtTimes font, html entities and modifications of ISO-8859-1 (Latin1)) were converted to Unicode/UTF-8 (ISO-10646) font encodings. This was enabled by the contributions to the Unicode standard made by the Medieval Unicode Font Initiative (MUFI), of which Wills is a member.

The indexing of skaldic diction (the figures known as kennings and heiti) has been automatic since the database’s inception. In this period Wills adapted the indexing to categorise the kennings and heiti in the database according to the system developed by Meissner, and to produce a facility for creating new analytical systems for these figures.

The use of a relational model for the project’s editions permitted the development of lexicographic resources in the database and Wills implemented such a system during this period. Contributors can link grammatical forms of words in the editions to dictionary headwords using an automated system which ‘learns’ the grammatical forms. In addition, variant manuscript forms can be added to the lexicon, as well as grammatical (parsing) information about every word.

Also added are facilities for producing manuscript stemmata, metrical analysis, plus searching of various materials in the corpus.

In 2012 Wills created a system for producing electronic links for in-text references in discursive materials for the editions (introductions, notes, etc.). This involved the automatic generation of text-matching strings for locating such references and the recording of references as two-way links. Users viewing such text will see automatic links to other parts of the database (stanzas, texts, etc.) and bibliographic resources. Where digitised pages are in the database or linked to the database, links from the page references will also be available. Viewing database and bibliographic items will also produce links back to the materials which reference them, creating an automatic index of all materials. As at May 2013, some 100,000 automatically-generated references can be produced in this way from the entered editions.

Also in 2012 volume I was successfully exported for print using an updated system and Unicode font encoding.

In 2013 Wills implemented considerable advances in the web interface, using the JQuery library for interactive features. These include the automatic animated reordering of verse and prose word orders for any stanza in the corpus, plus also advances in data entry. These have been particularly useful in overcoming the limitations of the relational data model for ordered data (linguistic or otherwise). Members of the project can now alter the order of words in the prose word order and translation through a drag-and-drop interface, as well as other materials that require ordered data, such as contents of volumes, stemmatic order of manuscripts and so on.

Some brief notes on the earlier history of the database:

• 1999: Initial interactive versions of Ragnarsdrápa produced using a Javascript interface
• May-June 2001: Tarrin starts creating a data structure for the information in the skaldic card index in Copenhagen
• Second half of 2001: Data is published on web using a perl interface and perl data structures
• Mid-2002: Detailed logging of changes is started
• Database develops from a tool for editors to the repository and publication system for the edition
• Database is converted to SQL format and an interface written in perl, including for updating the data
• Faster PHP interface developed; perl retained for various purposes, including filtering output for print version
• 2006-7: Volume VII produced from database
• Legacy perl code replaced by PHP scripts

References