In search applications, tuning relevance is a constant, iterative exercise to bring the right search results to your end users. The tooling in this repository aims to help the search relevance engineer and business user create the best search experience possible for application users without hiding internals from engineers who want to go deep into the details.
Behind the scenes, the plugin uses the Search Relevance OpenSearch plugin for resource management for each tool provided. For example, most use cases involve configuring and creating search configurations, query sets, and judgements. All of these resources are created, updated, deleted, and maintained by the Search Relevance plugin. When users are satisfied with the improvements to relevancy then they take the output and manually deploy the changes into their environment.
For tutorials on how to leverage these tools, see here.
Important
While shipping with OpenSearch, you must OPT IN to this feature. To enable this go to Dashboard Management -> advanced setting -> Search relevance -> turn on the toggle -> don’t forget to save.
The Query Comparison tool allows users to enter a plain text search query, run it with two different Query DSLs and compare the results side by side.
This tool lets you take a list of queries and compare them against two Search Configurations, and provides metrics that summarize the differences in results being returned by the two search configurations. Users can then drill in to the individual queries to eyeball the specific documents returned and how they have changed position.
Classic Information Retrieval is built around assessing search quality using Judgements, a rating of appropriateness between a query and a document returned. This tool lets you evaluate a Search Configuration and returns metrics like Average Precision, MAP, and NDCG.
Hybrid queries blend lexical and vector search results together. With the right Judgements, the Hybrid Optimizer will evaluate various ways of blending lexical and vector results together and provide an optimum balance.
See SECURITY for more information.
See CONTRIBUTING for more information.
This project is licensed under the Apache-2.0 License.