trustgraph/dev-tools/proc-group/groups/embeddings-store.yaml
Sunny Yang 2bdc930b2a
feat: hybrid retrieval (BM25 + vector RRF fusion) for document-RAG (#875) (#1030)
Adds a sparse keyword retrieval path beside the existing vector path in
document-RAG, fused by weighted Reciprocal Rank Fusion on chunk_id, behind
--retrieval-mode (vector | keyword | hybrid, default vector).

The keyword index is a new pluggable service (KeywordIndexService /
KeywordIndexClientSpec); the first backend is SQLite FTS5, consuming Chunk
messages off the ingestion stream and answering BM25 queries from one
process, since the index is a single local file. Query text is sanitized
into per-term quoted phrases (raw text is not valid FTS5 syntax), which
also makes dotted clause numbers and error codes exact-match without a
trigram index. Indexes are scoped per (workspace, collection) and dropped
on collection deletion.

The keyword-index client spec is only registered when the sparse path is
enabled, so existing flow definitions without keyword-index queues are
untouched; with retrieval_mode=vector the retrieval path is unchanged. In
hybrid mode a keyword-path failure degrades to vector-only.
2026-07-07 12:54:02 +01:00

53 lines
1.5 KiB
YAML

# Embeddings store. All Qdrant-backed vector query/write processors.
# One process owns the Qdrant driver pool for the whole group.
_defaults: &defaults
pubsub_backend: rabbitmq
rabbitmq_host: localhost
log_level: INFO
processors:
- class: trustgraph.query.doc_embeddings.qdrant.Processor
params:
<<: *defaults
id: doc-embeddings-query
store_uri: http://localhost:6333
- class: trustgraph.storage.doc_embeddings.qdrant.Processor
params:
<<: *defaults
id: doc-embeddings-write
store_uri: http://localhost:6333
- class: trustgraph.query.graph_embeddings.qdrant.Processor
params:
<<: *defaults
id: graph-embeddings-query
store_uri: http://localhost:6333
- class: trustgraph.storage.graph_embeddings.qdrant.Processor
params:
<<: *defaults
id: graph-embeddings-write
store_uri: http://localhost:6333
# Keyword (BM25) index: ingest-write and query in one processor, since
# the FTS5 index is a single local file.
- class: trustgraph.storage.kw_index.fts5.Processor
params:
<<: *defaults
id: kw-index
index_path: /tmp/tg-kw-index.db
- class: trustgraph.query.row_embeddings.qdrant.Processor
params:
<<: *defaults
id: row-embeddings-query
store_uri: http://localhost:6333
- class: trustgraph.storage.row_embeddings.qdrant.Processor
params:
<<: *defaults
id: row-embeddings-write
store_uri: http://localhost:6333