Posted On: Oct 11, 2023
Search Pipelines, a new feature in OpenSearch 2.9, make it easy to build query and result processing pipelines. This lets you build search query and result processing as a composition of modular processing steps without complicating your application software.
Individual processing steps can include query rewriters, natural language processors, result rerankers, and filters. Along with the search pipeline framework, OpenSearch now includes a variety of standard processors: a script processor, which allows you to preprocess search queries with Painless; a search request modifier which lets you add a filter in DSL; a field renamer; and so on. Search Pipelines are also the integration technology on which other important features are built, including score normalization for hybrid lexical and semantic searching, and AWS Personalize and AWS Kendra reranking. Pipelines can be explicitly invoked for the current query, or can be applied to all search requests on an index.
For information on upgrading to OpenSearch 2.9, please see documentation. OpenSearch 2.9 is now available in all AWS regions globally where Amazon OpenSearch service is available.
To learn more about Amazon OpenSearch Service, please visit the product page.