Feedback: Experimental Feature - Compare Search Results

In OpenSearch 2.4, we are introducing the Compare Search Results tool as an experimental feature.

The OpenSearch team is actively looking for your feedback and ideas to enhance our experimental features and capabilities. Please include a description of your use cases to help understand the context of your feedback.

3 Likes

is the team still looking for feedback @KrooshalUX?

If you’re interested, I offer some time at an upcoming community meeting to present / ask the community

Did this ever happen? I think we do have a few features to add to the comparison tool (images, summarization, better field display). We could use new GH issues to describe these features and then demo the current tool (or the one built for the ML playground) and review the issues.

This feature is useful, I would want this to be more collaborative. Currently, I don’t have a feature to save the session.

This is an amazing feature. Would love to be able to see the scores in the results, the total number of results (not only from the first page), and to export the results or to have the ability to share with others.

Thanks for the feedback! Great ideas. If you don’t see an issue related to one of your ideas here, would you be up for creating them?

This is the repo with all the code. Any of the issues with the [FEATURE] prefix would be good to review.GitHub - opensearch-project/dashboards-search-relevance: Tools to help search relevance engineers and business users tune search results for their OpenSearch applications.

I think export & share is related to [FEATURE] [Search Comparison Tool] Saved Configurations · Issue #284 · opensearch-project/dashboards-search-relevance · GitHub, so please let us know what you would like to see there.

It would be useful if compare search results works against an endpoint. So for example we could compare the results of a request from another search engine and also compare external reranking against the Kendra ranking and Personalised Search ranking plugins supported by Opensearch.