2.18.0.0

Version 2.18.0.0 Release Notes

Compatible with OpenSearch 2.18.0

Features

• Introduces ByFieldRerankProcessor for second level reranking on documents (#932)

Bug Fixes

• Fixed incorrect document order for nested aggregations in hybrid query (#956)

Enhancements

• Implement ignore_missing field in text chunking processors (#907)
• Added rescorer in hybrid query (#917)

2.17.0.0

Version 2.17.0.0 Release Notes

Compatible with OpenSearch 2.17.0

Enhancements

• Adds rescore parameter support (#885)

Bug Fixes

• Removing code to cut search results of hybrid search in the priority queue (#867)
• Fixed merge logic in hybrid query for multiple shards case (#877)

Infrastructure

• Update batch related tests to use batch_size in processor & refactor BWC version check (#852)

2.16.0.0

Version 2.16.0.0 Release Notes

Compatible with OpenSearch 2.16.0

Features

• Enable sorting and search_after features in Hybrid Search #827

Enhancements

• InferenceProcessor inherits from AbstractBatchingProcessor to support sub batching in processor #820
• Adds dynamic knn query parameters efsearch and nprobes #814
• Enable '.' for nested field in text embedding processor (#811)
• Enhance syntax for nested mapping in destination fields(#841)

Bug Fixes

• Fix function names and comments in the gradle file for BWC tests (#795)
• Fix for missing HybridQuery results when concurrent segment search is enabled (#800)

Infrastructure

• Add BWC for batch ingestion (#769)
• Add backward test cases for neural sparse two phase processor (#777)
• Fix CI for JDK upgrade towards 21 (#835)
• Maven publishing workflow by upgrade jdk to 21 (#837)

2.15.0.0

Version 2.15.0.0 Release Notes

Compatible with OpenSearch 2.15.0

Features

• Speed up NeuralSparseQuery by two-phase using a custom search pipeline.(#646)
• Support batchExecute in TextEmbeddingProcessor and SparseEncodingProcessor (#743)

Enhancements

• Pass empty doc collector instead of top docs collector to improve hybrid query latencies by 20% (#731)
• Optimize parameter parsing in text chunking processor (#733)
• Use lazy initialization for priority queue of hits and scores to improve latencies by 20% (#746)
• Optimize max score calculation in the Query Phase of the Hybrid Search (765)
• Implement parallel execution of sub-queries for hybrid search (#749)

Bug Fixes

• Total hit count fix in Hybrid Query (756)
• Fix map type validation issue in multiple pipeline processors (#661)

Infrastructure

• Disable memory circuit breaker for integ tests (#770)

2.14.0.0

Version 2.14.0.0 Release Notes

Compatible with OpenSearch 2.14.0

Features

• Support k-NN radial search parameters in neural search(#697)

Enhancements

• BWC tests for text chunking processor (#661)
• Add support for request_cache flag in hybrid query (#663)
• Allowing execution of hybrid query on index alias with filters (#670)
• Allowing query by raw tokens in neural_sparse query (#693)
• Removed stream.findFirst implementation to use more native iteration implement to improve hybrid query latencies by 35% (#706)
• Removed map of subquery to subquery index in favor of storing index as part of disi wrapper to improve hybrid query latencies by 20% (#711)
• Avoid change max_chunk_limit exceed exception in text chunking processor (#717)

Bug Fixes

• Fix async actions are left in neural_sparse query (#438)
• Fix typo for sparse encoding processor factory(#578)
• Add non-null check for queryBuilder in NeuralQueryEnricherProcessor (#615)
• Add max_token_score field placeholder in NeuralSparseQueryBuilder to fix the rolling-upgrade from 2.x nodes bwc tests. (#696)
• Fix multi node "no such index" error in text chunking processor. (#713)

Infrastructure

• Adding integration tests for scenario of hybrid query with aggregations (#632)

Maintenance

• Update bwc tests for neural_query_enricher neural_sparse search (#652)

2.13.0.0

Version 2.13.0.0 Release Notes

Compatible with OpenSearch 2.13.0

Features

• Implement document chunking processor with fixed token length and delimiter algorithm (#607)
• Enabled support for applying default modelId in neural sparse query (#614

Enhancements

• Adding aggregations in hybrid query (#630)
• Support for post filter in hybrid query (#633)

Bug Fixes

• Fix typo for sparse encoding processor factory(#600)
• Add non-null check for queryBuilder in NeuralQueryEnricherProcessor (#619)
• Fix runtime exceptions in hybrid query for case when sub-query scorer return TwoPhase iterator that is incompatible with DISI iterator (#624)

2.12.0.0

Version 2.12.0.0 Release Notes

Compatible with OpenSearch 2.12.0

Features

• Add rerank processor interface and ml-commons reranker (#494)

Bug Fixes

• Fixed exception for case when Hybrid query being wrapped into bool query (#490)
• Hybrid query and nested type fields (#498)
• Fixing multiple issues reported in #497 (#524)
• Fix Flaky test reported in #433 (#533)
• Enable support for default model id on HybridQueryBuilder (#541)
• Fix Flaky test reported in #384 (#559)
• Add validations for reranker requests per #555 (#562)

Infrastructure

• BWC tests for Neural Search (#515)
• Github action to run integ tests in secure opensearch cluster (#535)
• BWC tests for Multimodal search, Hybrid Search and Neural Sparse Search (#533)
• Distribution bundle bwc tests ([#579])(#579)

Maintenance

• Added support for jdk-21 (#500))
• Update spotless and eclipse dependencies (#589)

Refactoring

• Deprecate the max_token_score field in neural_sparse query clause (#478)
• Added spotless check in the build (#515)

2.11.0.0

Version 2.11.0.0 Release Notes

Compatible with OpenSearch 2.11.0

Features

• Support sparse semantic retrieval by introducing sparse_encoding ingest processor and query builder (#333)
• Enabled support for applying default modelId in neural search query (#337
• Added Multimodal semantic search feature (#359)

Enhancements

• Add max_token_score parameter to improve the execution efficiency for neural_sparse query clause (#348)

Bug Fixes

• Fixed exception in Hybrid Query for one shard and multiple node (#396)

Maintenance

• Consumed latest changes from core, use QueryPhaseSearcherWrapper as parent class for Hybrid QPS (#356)

2.10.0.0

Version 2.10.0.0 Release Notes

Compatible with OpenSearch 2.10.0

Features

• Improved Hybrid Search relevancy by Score Normalization and Combination (#241)

Enhancements

• Changed format for hybrid query results to a single list of scores with delimiter (#259)
• Added validations for score combination weights in Hybrid Search (#265)
• Made hybrid search active by default (#274)

2.9.0.0

Version 2.9.0.0 Release Notes

Compatible with OpenSearch 2.9.0

Maintenance

Increment version to 2.9.0-SNAPSHOT (#191)

Bug Fixes

Fix update document with knnn_vector size not matching issue (#208)