28 lines
1.3 KiB
SQL
28 lines
1.3 KiB
SQL
-- Migration: Add pgvector support to icd_codes table
|
|
|
|
-- Enable pgvector extension
|
|
CREATE EXTENSION IF NOT EXISTS vector;
|
|
|
|
-- Add embedding column with pgvector type
|
|
ALTER TABLE "icd_codes" ADD COLUMN IF NOT EXISTS "embedding" vector(1536);
|
|
|
|
-- Add metadata column for LangChain pgvector
|
|
ALTER TABLE "icd_codes" ADD COLUMN IF NOT EXISTS "metadata" JSONB;
|
|
|
|
-- Add content column for LangChain pgvector
|
|
ALTER TABLE "icd_codes" ADD COLUMN IF NOT EXISTS "content" TEXT;
|
|
|
|
-- Create ivfflat index for fast vector similarity search
|
|
CREATE INDEX IF NOT EXISTS "icd_codes_embedding_idx" ON "icd_codes"
|
|
USING ivfflat (embedding vector_cosine_ops) WITH (lists = 100);
|
|
|
|
-- Create index on metadata for fast JSON queries
|
|
CREATE INDEX IF NOT EXISTS "icd_codes_metadata_idx" ON "icd_codes" USING GIN (metadata);
|
|
|
|
-- Add comments for documentation
|
|
COMMENT ON COLUMN "icd_codes"."embedding" IS 'Vector embedding for semantic search using pgvector (1536 dimensions)';
|
|
COMMENT ON COLUMN "icd_codes"."metadata" IS 'JSON metadata for LangChain pgvector operations';
|
|
COMMENT ON COLUMN "icd_codes"."content" IS 'Text content for LangChain pgvector operations';
|
|
COMMENT ON INDEX "icd_codes_embedding_idx" IS 'IVFFlat index for fast cosine similarity search with 100 lists';
|
|
COMMENT ON INDEX "icd_codes_metadata_idx" IS 'GIN index for fast JSON metadata queries';
|