ideas on Question Engine architecture

[David]

The UI is wicket based application, possibly not OSGi. Ideally it
would be clusterable at some point but not necessary at the outset.
It stores minimal user info:
authentication info
email
ID of the user's data server

The UI performs these functions:
new accounts
receive payments, if/when applicable
authenticate users
change account info
communicates with the user's data server to pull personalized data
(questions, advisories, data)
render questions, personalized info

The data servers are OSGi applications, using Eclipse ECF to
communicate with the UI server. They store data in an embedded Jena
TDB database. They are capable of applying forward-chained rules by
doing SPARQL CONSTRUCT queries.
To the UI application, they provide lookup functions:
get latest questions for the user
get latest advisories for the user
pull any user data for charting or other purposes

The data servers also provide these background functions: mining a
user's data, in search of new association rules.

One or more additional learning/INQLE servers are used to do data
mining across all of the data servers, in search of findings across
all participants.

Finally, rule authors will need a system for authoring rules.
Ideally, this would be an OSGi application which is similar to the
main UI, but with added rule authoring functionality.