I think you're kinda missing the idea of index time boosting. The semantic of this (as I remember Chris Hostetter explaining) is "this document's content is more important than other document's content".
By doing an index-time boost that's the same for all your documents, you're effectively doing nothing to the relative ranks of the results. Not quite sure what Luke is doing here, but using &debugQuery=on will give you the actual scores of the actual documents. And if you're doing anything like wildcards or *:* queries, shortcuts are taken that set the scores to 1.0. If none of that helps, I'm out of my depth <G>.. Best Erick On Wed, Apr 24, 2013 at 6:01 AM, Timothy Hill <timothy.d.h...@gmail.com> wrote: > Hello, all > > I have recently been attempting to apply index-time boosts to fields using > the following syntax: > > <add> > <doc> > <field name="important_field" boost="5">bleah bleah bleah</field> > <field name="standard_field" boost="2">content here</field> > <field name="trivial_field">content here</field> > </doc> > <doc> > <field name="important_field" boost="5">content here</field> > <field name="standard_field" boost="2">bleah bleah bleah</field> > <field name="trivial_field">content here</field> > </doc> > </add> > > The intention is that matches on important_field should be more important > to score than matches on trivial_field (so that a search across all fields > for the term 'content' would return the second document above the first), > while still being able to use the standard query parser. > > Looking at output from Luke, however, all fields are reported as having a > boost of 1.0. > > The following possibilities occur to me. > > (1) The entire index-time-boosting approach is misconceived > (2) Luke is misreporting, because index-time boosting alters more > fundamental aspects of scoring (tf-idf calculations, I suppose), and the > index-time boost is thus invisible to it > (3) Some combination of (1) and (2) > > Can anyone help illuminate the situation for me? Documentation for these > questions seems patchy. > > Thanks, > > Tim
