OpenSearch Learning to Rank plugin

During this evening meetup, It was requested to port Elasticsearch Learning to Rank plugin to Opendistro.

I published a fork update to 7.10 version and initial working binaries at opensearch-learning-to-rank.

Only 14 tests or above ~200 are not working due to securemock issues.

Any feedback is welcomed.


Wow. That was fast. Did you use the automated tool you spoke about in the meeting or was this done by hand?

The big part was done by the script to convert names and classes to opensearch namespaces. It’s not perfect but it does the majority of the change.
Elasticsearch code and opensearch code are quite the same.
To compile the code, I had to remove some references to old Elasticsearch version (7.7, 6.7, …).
And then after the testing, it’s quite fast to do a github release :wink:


hi, we are currently evaluating open-search and the ltr plugin (using the fork for the same) Installed open-search and the ltr plugin. Now trying write the code which would use the ltr apis to search (not normal search, that I have already done), but using ltr based search. Has any one tried some thing on these lines (hello-ltr uses elastic search sever), want to use the same for opensearch server

Hey @anshuk check out Grant Ingersoll’s repo for some good examples.

Also, take a look at OpenSearch - Hello LTR Update by macohen · Pull Request #91 · o19s/hello-ltr · GitHub; it’s nowhere near complete, but uses Grant’s plugin. We have a propsal to move that repo into the opensearch-project as well. [PROPOSAL] Make the Opensearch LTR Plugin a part of the OpenSearch official plugin ecosystem · Issue #217 · opensearch-project/opensearch-plugins · GitHub.