I want to implement some pseudo-relevance feedback (something like: Relevance Based Language Model - Lavrenko, Croft, SIGIR 2001) mechanism in the Elasticsearch. I already have an implementation in Lucene, tested with TREC data (https://github.com/dwaipayanroy/Relevance-Model). Is there a way to do that in the Elasticsearch as well?
I want to incorporate the relevance feedback inside a running system which is build with Elasticsearch.
Any help will be appreciated.