The scenario is as follows:
1.) created an index with documents.
2.) tried to execute more_like_this query on the index. It tries to search a keyword in a particular field in all the indexed docs.
3.) the obtained result contains unexpected result.
4.) the keyword has two letters say A and B, ("A B").
5.) the first ranked document contains B and has 4 words in total in the field (based on which search is performed).
6.) the second ranked document contains B followed by A ("B A") and the length of the field is 15 words long.
7.) the above result seems incorrect. The more relevant document got lesser relevance score than the lesser relevant document. I have a feeling that fieldnorm being enabled might be the reason.
Disabling fieldnorm is not working, hence can't narrow down as of now. I've created a post for it as well in the forum. Can there be any other possible reasons for this issue?
Kindly let me know if any more clarification is required on the issue.