[Developers] MQL and Metaweb data model

[Michael J Cafarella]

Hi Scott,

Thanks for your detailed reply.  I really appreciate the time you took.

It sounds like your database is actually more similar to a search engine's 
document-store than to a relational database.

In a search engine, documents are "identities" which have a bag of 
associated values.  You could imagine marking a document with text that 
says "band=The_Police" or "type=Album".  You could retrieve results for 
either the "band" query or the "type" query easily.

The "wide-semijoin" that I described would entail looking up a number of 
these different sets, sorting them, then performing an intersection-style 
operation between them.  Like you say, this would be difficult eventually 
(say, 1000 different types) but you might not run into problems for 
any real-world queries.

This is very interesting stuff.  I hope you consider publishing an article 
about it someday.

--Mike