I'm building a search engine for which I already know the weights of every term. I'd like to be able to specify these weights at document insertion time, replacing TF-IDF. Is it possible to overwrite the term_vector?
Surfing around I see solutions using payloads, but those also need a function_score & are slow.
I ended up using the delimited payload token filter example from here:
combined with limiting the script to a rescore, for performance reasons. This is a lot harder than it ought to be! The completion suggester is a good existing example of how to allow easy specification of ranking info in Elastic.
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.