Distributed reasoning...
System was distributed using Java & XML
Server uses relaxation to produce reasonable number of items, e.g. a few hundred
Pushed to client side applet via XML
- runs simple nearest neighbour algorithm to find closest set
- Simply minimise similarity measure
- ?i f(Ti,Si) wi
- where summation over features i
- f(Ti,Si) difference measure on feature i between cases S, T
- wi is weight of feature i
- obtain full details of closest set by ftp