According to http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/search-suggesters-phrase.htmlraising
prefix_len should improve performance and lowering it will make
things slower. If I set it to 0, is that really going to kill
performance? I've got someone complaining that "chokolade" doesn't correct
to "schokolade". As a someone who's highschool German faded long ago from
my memory I'd pronounce both words the same way.
Nick I think you will suffer much from setting it to 0 but give it a try.
To do what you want to do have a second candidate generator that uses a
reverse filter with a reasonable prefix len?
simon
On Friday, October 18, 2013 4:05:15 PM UTC+2, Nikolas Everett wrote:
According to Elasticsearch Platform — Find real-time answers at scale | Elastic prefix_len should improve performance and lowering it will make
things slower. If I set it to 0, is that really going to kill
performance? I've got someone complaining that "chokolade" doesn't correct
to "schokolade". As a someone who's highschool German faded long ago from
my memory I'd pronounce both words the same way.
I'm not sure I'll get a chance to work on this soon but I'll remember to
try the reverse filters when I can. Is the goal of the prefix len just to
cut down on the number of candidates or to improve the locality of lookups
in the term statistics? Actually, I imagine both.
Nick I think you will suffer much from setting it to 0 but give it a try.
To do what you want to do have a second candidate generator that uses a
reverse filter with a reasonable prefix len?
simon
On Friday, October 18, 2013 4:05:15 PM UTC+2, Nikolas Everett wrote:
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.