: I have explored ranking formula. As far as I understand, it : seems query-boost value is used only in "queryNorm" (to be exact : in : "sumOfSquaredWeights" ) that too inversely.
No, the queryNorm is a balancer that takes into account all of the boosts of all of the constituante parts of a complex query, but for each primitive query the boost is used directly... http://lucene.apache.org/java/2_9_0/api/core/org/apache/lucene/search/Similarity.html#formula_termBoost ...that example refers to it as "t.getBoost()" because it's typically a boost applied to a TermQuery, but the same holds true for any primitive query. : I am just wondering if it has any direct effect on ranking : score. Like, more queryBoost resulting in more score. But here as it absolutely -- but only within the context of scores for other docs in the same query. If your query is... A^33333 B^222 C^11 ...then matches on clause "A" will contribute to the final score much greater then matches to clause B or clause C; but the final score values for matches won't neccessarily be in in a higher range then the scores for the same docs using a query like... A^3 B^2 C^1 What matters is that the relative order will be very different. -Hoss
