Data Warehousing is now a well-established part of the business and scientific worlds. However, up until recently, data warehouses were restricted to modeling essentially numerical data – examples being sales figures in the business arena (in say Wal-Mart’s data warehouse (Westerman, 2000)) and astronomical data (for example SKICAT) in scientific research, with textual data providing a descriptive rather than a central analytic role. The lack of ability of data warehouses to cope with mainly non-numeric data is particularly problematic for humanities1 research utilizing material such as memoirs and trade directories. Recent innovations have opened up possibilities for ‘non-numeric’ data warehouses, making them widely accessible to humanities research for the first time. Due to its irregular and complex nature, humanities research data is often difficult to model, and manipulating time shifts in a relational database is problematic as is fitting such data into a normalized data model. History and linguistics are exemplars of areas where relational databases are cumbersome and which would benefit from the greater freedom afforded by data warehouse dimensional modeling.
TopBackground
Hudson (2001, p. 240) declared relational databases to be the predominant software used in recent, historical research involving computing. Historical databases have been created using different types of data from diverse countries and time periods. Some databases are modest and independent, others part of a larger conglomerate like the North Atlantic Population Project (NAPP) project that entails integrating international census data. One issue that is essential to good database creation is data modeling; which has been contentiously debated recently in historical circles.
When reviewing relational modeling in historical research, (Bradley, 1994) contrasted ‘straightforward’ business data with incomplete, irregular, complex- or semi-structured historical data. He noted that the relational model worked well for simply-structured business data, but could be tortuous to use for historical data. (Breure, 1995) pointed out the advantages of inputting data into a model that matches it closely, something that is very hard to achieve with the relational model. (Burt2 and James, 1996) considered the relative freedom of using source-oriented data modeling (Denley, 1994) as compared to relational modeling with its restrictions due to normalization (which splits data into many separate tables), and highlighted the possibilities of data warehouses. Normalization is not the only hurdle historians encounter when using the relational model.
Date and time fields provide particular difficulties: historical dating systems encompass a number of different calendars, including the Western, Islamic, Revolutionary and Byzantine. Historical data may refer to ‘the first Sunday after Michaelmas’, requiring calculation before a date may be entered into a database. Unfortunately, some databases and spreadsheets cannot handle dates falling outside the late 20th century. Similarly, for researchers in historical geography, it might be necessary to calculate dates based on the local introduction of the Gregorian calendar, for example. These difficulties can be time-consuming and arduous for researchers. Awkward and irregular data with abstruse dating systems thus do not fit easily into a relational model that does not lend itself to hierarchical data. Many of these problems also occur in linguistics computing.
Linguistics is a data-rich field, with multifarious forms for words, multitudinous rules for coding sounds, words and phrases, and also numerous other parameters - geography, educational and social status. Databases are used for housing many types of linguistic data from a variety of research domains - phonetics, phonology, morphology, syntax, lexicography, computer-assisted learning (CAL), historical linguistics and dialectology. Data integrity and consistency are of utmost importance in this field. Relational DataBase Management Systems (RDBMSs) are able to provide this, together with powerful and flexible search facilities (Nerbonne, 1998, introduction).