What does TOP mean ?

[Markus Demleitner]

Hi Petr, hi SSA crowd,

On Sat, Apr 25, 2015 at 01:38:26AM +0200, Petr Skoda wrote:
> 
> And does anybody implemented TOP in scoring manner in SSAP ?

Well, DaCHS accepts TOP but treats it equivalently to MAXREC so far
(which I claim is legal, as that simply means I give all matching
results the same score).

The basic difficulty in doing a meaningful implementation is that you
need to compute a score, and this kind of scoring in general is
either difficult (because to do it properly you need a statistical
model of both the data and the user) or haphazard (involving
combining more or less ad-hoc measures for how good a match is for a
certain constraint weighted... somehow).

I think it would be a moderately worthwhile effort to create a model
of either kind for SSAP's parameter set and validate it with user
studies.  Has anyone perhaps already started on such a thing?

My suspicion is that this model could work well across a multitude of
services and domains with no or just a single parameter.  Be that as
it may, I'm afraid I don't see GAVO tackling something like that any
time soon.

Cheers,

        Markus