feat(retrieval): add LLM query expansion with multi-query fusion#159
Closed
Salomondiei08 wants to merge 580 commits into
Closed
feat(retrieval): add LLM query expansion with multi-query fusion#159Salomondiei08 wants to merge 580 commits into
Salomondiei08 wants to merge 580 commits into
Conversation
feat:support eventlog/foresight in keyword/vector/hybird search mode for v1-api See merge request npc-work/aic/ai/evermemos-opensource!52
Resolved conflicts: - demo/extract_memory.py: Keep V3 API config and prompt_clear_data function - Updated type imports: Memory -> BaseMemory, ForesightItem -> Foresight - Kept PromptManager architecture with get_prompt_by() - Used absolute imports per project standards Risk check passed: all imports, prompts, and extractor compatibility verified.
feat: Optimize demo and related memory chat logic See merge request npc-work/aic/ai/evermemos-opensource!40
…x MEMORY_LANGUAGE for evaluation
:global_with_meridians: utc :global_with_meridians: utc :global_with_meridians: utc
:global_with_meridians: utc See merge request npc-work/aic/ai/evermemos-opensource!55
refactor:v1 api search, support rrf and agentic search, update readme See merge request npc-work/aic/ai/evermemos-opensource!53
Feature/boundary opt See merge request npc-work/aic/ai/evermemos-opensource!54
- Remove duplicate summary_text assignment in conv_memcell_extractor.py - Remove redundant cluster state file exports in stage1_memcells_extraction.py since cluster_storage.save_cluster_state already handles saving both cluster_state and cluster_map internally Co-authored-by: root <root@nlu-guishangtong-sbajo4-0.nlu-guishangtong-sbajo4.lizr-a.svc.dev00-bcebj.local> Co-authored-by: Claude <noreply@anthropic.com>
* refactor: normalize request converter parsing * refactor: dedupe request param parsing in memory controller * refactor: harden lightweight retrieval similarity handling * fix: close agentic services during business shutdown
* docs: update README with community links and branding * docs: update community links and badge labels
* docs: update README with new competition banner and examples * docs: update light mode divider image asset link * docs: update README with corrected spelling and added links * docs: update README with competition and example links * docs: remove commented section and fix light mode divider * docs: remove divider images from README
…Mind-AI#116) Update the OpenClaw Long-Term Memory Plugin section to reflect its release status by removing "coming this week" label. Also fix the markdown image link syntax to properly wrap the image in a link to the agent_memory branch.
…#119) Update the link for the use case image and add reference to the EverMind + OpenClaw plugin repository. The change clarifies the project structure and provides direct access to the plugin code.
* docs: update README with corrected links and plugin info Update the link for the use case image and add reference to the EverMind + OpenClaw plugin repository. The change clarifies the project structure and provides direct access to the plugin code. * docs: update usecases image link to point to agent_memory branch
- Introduced the EverOS OpenClaw Plugin, enabling full-lifecycle memory management for OpenClaw through the EverOS backend. - Added core files including index.js for plugin registration, openclaw.plugin.json for metadata, and README documentation. - Implemented installation script for easy setup and configuration. - Included support for memory retrieval and storage via the EverMemOS API. - Added Chinese translation of the README for broader accessibility.
…Plugin - Updated the searchMemories function to consolidate search requests and handle profile fetching more efficiently. - Removed unsupported memory types from configuration and adjusted memory type handling in search logic. - Enhanced message processing by stripping channel metadata and including pending messages in responses. - Improved parsing of search responses to accommodate new profile data structures and pending messages.
* chore: rename project from EverMemOS to EverOS Update all references to the project name, including: - NOTICE and project metadata files - Documentation and README files - Code comments and docstrings - Configuration files and system names - Example data and demo references - Adapter directories and imports - Docker container names and labels - Security report subject line - Plugin descriptions and metadata This rebranding consolidates the project under the EverOS name while maintaining full compatibility. * docs: update README.zh.md with new project focus and structure Update the Chinese README to reflect the project's current focus on Memory Sparse Attention (MSA) instead of the Memory Genesis Competition. - Replace competition announcement with MSA paper and technical highlights. - Simplify and update the table of contents. - Reorder and comment out old badge sections. - Update banner image and arXiv badge link. - General cleanup and formatting improvements.
…)" (EverMind-AI#152) This reverts commit edff40a.
…from EverMemOS to EverOS
Addresses vocabulary mismatch failure mode in hybrid retrieval where stored memories use different terms than the query (e.g. 'rescue inhaler protocol' stored vs 'gym bag' queried). Changes: - src/memory_layer/query_expansion.py: new module implementing expand_query() which generates 2-3 LLM paraphrase variants of a query using a structured prompt. Falls back silently (returns []) on any LLM failure. Also provides merge_hits_by_id() for union-dedup of per-query result sets. - src/agentic_layer/memory_manager.py: retrieve_mem_hybrid() now routes through _search_hybrid_with_query_expansion() instead of _search_hybrid(). This runs keyword+vector search in parallel for the original query and each variant, merges results by memory id (original-query scores preserved), then reranks the union against the original query for consistent scoring. Falls back to plain hybrid search when expansion produces no variants. Inspired by RAG-Fusion / HyDE. Recall improvement is measurable via Recall@K with single-query vs expanded-query baselines.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
11 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Checklist
Additional Notes
_search_hybridis unchanged and still used by the AGENTIC and RRF retrieval paths — only the HYBRID path is affectedBreaking Changes
None.
_search_hybridis preserved as-is. The new routing inretrieve_mem_hybridis fully backward compatible and falls back to the original behavior on any failure.