Blog
-
You can now see ML model status in OpenSearch Dashboards
Before OpenSearch 2.6, checking the status of a deployed machine learning (ML) model in OpenSearch could only be done by using the ML Commons Profile API. Although the Profile API returned data organized by ML nodes, it could not tell you the status of a specific model deployed on that node.
Read More -
Technical roadmap: OpenSearch extensibility
The OpenSearch plugin framework presents a number of challenges for users and developers. To solve these, we've embarked on a journey to replace the plugin mechanism with extensions.
Read More -
Meet your new OpenSearch Project Slack workspace
The OpenSearch Project is happy to announce a new Slack community workspace to provide open communication channels for anyone interested in OpenSearch.
Read More -
The ABCs of semantic search in OpenSearch: Architectures, benchmarks, and combination strategies
In an earlier blog post, we described different ways of building a semantic search engine in OpenSearch. In this post, we'll dive further into the science behind it. We'll discuss the benefits of combining keyword-based search with neural search, the architecture and model options, and benchmarking tests and results. First, we'll provide an overview of our proposed solutions and a...
Read More -
Expanding k-NN with Lucene approximate nearest neighbor search
OpenSearch pioneered k-nearest neighbor (k-NN) within search engines in 2019, and developers have adopted it enthusiastically on sets of millions or even billions of vectors. OpenSearch continues to innovate in the area of k-NN support. OpenSearch 2.2 added the Lucene 9.0 implementation of k-NN, and OpenSearch 2.4 added adaptive filtering.
Read More