[DDI-users] Storing XML

[Joachim Wackerow]

Hi I-Lin,

I am interested in how you use Oracle for storing the DDIs:

- Do you store the whole DDI file of a study in one XMLType field?
- Is it possible to update only some DDI fields inside the XMLType?
- Is the XPath capability of Oracle enough for your needs?
- How is the performance doing queries to the XMLType?

Together with colleagues of the German Microdata Lab (ZUMA) I am working 
on a information system on microdata (MISSY). The system should provide 
information (not the data) on the German Microcensus (a yearly census of 
1% of the population from the Federal Statistics Bureau) for interested 
social science researchers.

The first plan was to store the DDI-compatible information in a 
relational database. In a first prototype we made a mapping of the 
database schema to some DDI tags. But a mapping of the whole DDI to the 
database schema seems to be very complicated, and very unefficient in 
transferring the information between the both data models. So we are 
considering to store the DDI XML files in a XML enabled database or in a 
native XML database. At ZUMA we have also Oracle.

Generally the XML native databases are still in a fast development 
process. XQuery and XUpdate implementations are still rare. As I know, 
Oracle hasn't XQuery or XUpdate capabilities.

Regards, Achim

I-Lin Kuo wrote:
> Hi Andrew,
> 
> Here at the Inter-University Consortium for Political and Social 
> Research (ICPSR), we're converting all our existing codebooks into DDI 
> format and developing a database to enable us to search for variables 
> across all the different survey codebook DDIs. We are using Oracle 9i.
> 
> The choice of Oracle 9i is due to the fact that we have a site license, 
> but nonetheless, it has pretty good XML capabilities (Oracle 8i and SQL 
> Server are far behind Oracle 9i in this respect, and would not have been 
> suitable for our needs), including:
> 
>         - limited XPath capability
>         - storing XML fragments in an Oracle native XML format (XMLType)
>         - full-text indexing within a field of XMLType
>         - output directly into XML or XSLT via Oracle's XSQL
>         - ability to store XML directly using object technology without 
> having to map tags and attributes to database fields
> 
> I haven't looked at Xindice or Tamino, because 9i was good enough for us.
>