Looks like a multi-field facet counting matches rather than documents
resulting in double-counting documents. Is it possible to control this
behavior? Wold be nice to have a flag in facet def. to select desired
behavior - it could be very confusing in some use cases to have more
matches on a facet than you have documents.
Looks like a multi-field facet counting matches rather than documents
resulting in double-counting documents. Is it possible to control this
behavior? Wold be nice to have a flag in facet def. to select desired
behavior - it could be very confusing in some use cases to have more
matches on a facet than you have documents.
Thank you Ivan but wouldn't I get a big hit on performance? Simple term
facet I assume is evaluated using inverted index while script based faced
will be evaluated by loading and faceting every document. In any case
ability to double or not-double counting would be a very helpful feature to
have. I would say not double-counting seems to be a more common need then
the opposite
On Tue, Jan 22, 2013 at 5:18 PM, Ivan Brusic ivan@brusic.com wrote:
You can use a script field with the terms facet where the script is
the aggregation of the two fields.
Looks like a multi-field facet counting matches rather than documents
resulting in double-counting documents. Is it possible to control this
behavior? Wold be nice to have a flag in facet def. to select desired
behavior - it could be very confusing in some use cases to have more
matches on a facet than you have documents.
That's odd. I tought a facet count was always made per doc.
Could it be that your facet is a nested facet? In witch case the count
would be made by sub-doc, thus resulting in a total greater than your
number of docs. (If I am not mistaken)
Le mardi 22 janvier 2013 22:31:57 UTC+1, AlexR a écrit :
Looks like a multi-field facet counting matches rather than documents
resulting in double-counting documents. Is it possible to control this
behavior? Wold be nice to have a flag in facet def. to select desired
behavior - it could be very confusing in some use cases to have more
matches on a facet than you have documents.
Actually I do not know if it does. I am very new to ES. I just assumed it takes number of top teems and finds counts for them because queries with large result set did mot seem ti be much slower than smal ones. I need to eead up on lucine stoeage organization...
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.