NTS:: Some ideas for a Windows-friendly approach for basic indexing of OAI-PMH content (amongst other sources).

Thinking about my brief flirtation with an EAV variant, I've realised that I should probably stick with a traditional structure, (and possibly use views to produce the output I would like).

1. Grab OAI-PMH content


2. For each record from OAI source, store it in Postgres (temp table? - alternatively use a table for each OAI source)


3. Create Postgres indices/tables from xpath queries... a standard xpath query wouldn't work with all OAI sources, (some include non Dublin Core content such as DIDL)
   


4. Run Sphinx over select Postgres tables


5. (Empty the Postgres temp table?)

Then can use Sphinx for other sources as well...

NTS:: Database structure and "flattened MaRC"

Thinking about how MaRC works, makes me think that subject (record?), property, value is not enough... or maybe I've not caught on to RDF just yet?

Column structure for a simple database (along the lines of XML and a flattened MaRC structure):

(okay, so this is not going to set the world on fire, but I think it's useable)

ROW # (autoincrement) | RECORD | LINE | ORDER | VALUE
1 | 1 | creator:surname | 001 | Smith
2 | 1 | creator:firstname | 001 | James
3 | 1 | title:entire | 001 | MySQL : Exploring database design
4 | 1 | published:date | 001 | 2004
5 | 1 | published:edition | 001 | [1st]
6 | 1 | published:date | 002 | 2006
7 | 1 | published:edition | 002 | 2nd