lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Nicolás Lichtmaier <nicol...@wolfram.com.INVALID>
Subject Conceptual doubt about "feature fields" with regards to addition/multiplication of scores
Date Mon, 01 Mar 2021 14:36:08 GMT
TL;DR: Why score addition is recommended in examples when using these? 
Doesn't that make the resulting score sensible to the number of boolean 
clauses?

I've been reading about feature fields (in particular the "dynamic" 
feature field you get with LongPoint.newDistanceFeatureQuery(), for 
adding a "recency" factor), and I see the recommended usage is to *add* 
that query to a boolean, like this: termquery1 should(featureQuery).

As BooleanQuery adds the score of the clauses in other queries the 
infuence of the clauses will vary, e.g.: If the query is: termquery1 
termquery2 should(featureQuery) .... then the score contribution of the 
feature field will be more diluted and the recency will be less 
relevant. That seems wrong. Wouldn't it be more appropriate to multiply 
the entirety of the score by a "recency" factor (got from the "feature 
field query")?

How this recency thing is normally implemented?

Thanks!


---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org


Mime
View raw message