From many to one: converting different databases into an XML database

Article type
Authors
Ekqvist M, Wahlbeck K
Abstract
Background: The PsiTri web service (psitri.stakes.fi) and its background database are formed by integrating separate trial databases from the Cochrane Mental Health Review Groups (CRG). Three of the CRG groups use a so-called flat-type database (ProCite), whilst two others use a relational-type database (MeerKat).

Objectives: The aim is to easily and quickly integrate the groupsÂ’ databases to make them publishable as Web-services. The monthly updated databases must be integrated and converted smoothly and efficiently and published to the Web, the process taking a maximum of two working days.

Methods: The most challenging part of the data processing is to integrate different kinds of databases into one uniform relational database and then to convert it into the Extensible Markup Language (XML). Database integration is currently managed with the help of routines created with MS Access (Figure 1). At the first stage, databases are transferred by the Cochrane Mental Health Review Groups (CRG) to the FTP server.
At the second stage, separate databases are saved in group-specific folders. The ProCite databases are exported into tabulator separated text files and then connected to the Access-database format.
At the third stage the Access-databases of the ProCite groups are combined with the existing MeerKat databases. The whole process is managed by MS Access, with execution orders controlled either by MS Access macros or Visual Basic for Application (VBA) modules.
At the fourth stage, the trial databases are combined in the correct field order within the so-called collecting table from which the final XML file is produced.

Results: The integration systems and conversation to the XML database makes it a flexible way to update PsiTri web-services whenever new data has been added to the source databases, whilst also allowing for an easier implementation of new techniques.

Conclusions: The database integration system is a quick and efficient way to publish separately updated databases in a web-based environment several times a year. The conversion into XML considerably shortens the publication time. A platform independent system will also allow the use of portable data terminal equipment. The MS Access background database can currently handle up to 40 000 trials.