From 0ed8f140153dcf58d2754db6a59d7d6116f92772 Mon Sep 17 00:00:00 2001 From: Weves Date: Wed, 15 Nov 2023 14:26:00 -0800 Subject: [PATCH] Improve Vespa filtering performance --- .../document_index/vespa/app_config/schemas/danswer_chunk.sd | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/backend/danswer/document_index/vespa/app_config/schemas/danswer_chunk.sd b/backend/danswer/document_index/vespa/app_config/schemas/danswer_chunk.sd index 7573e5219..8d68fb497 100644 --- a/backend/danswer/document_index/vespa/app_config/schemas/danswer_chunk.sd +++ b/backend/danswer/document_index/vespa/app_config/schemas/danswer_chunk.sd @@ -31,6 +31,8 @@ schema danswer_chunk { # https://docs.vespa.ai/en/attributes.html potential enum store for speed, but probably not worth it field source_type type string { indexing: summary | attribute + rank: filter + attribute: fast-search } # Can also index links https://docs.vespa.ai/en/reference/schema-reference.html#attribute # URL type matching @@ -61,6 +63,7 @@ schema danswer_chunk { } field hidden type bool { indexing: summary | attribute + rank: filter } field metadata type string { indexing: summary | attribute @@ -82,10 +85,12 @@ schema danswer_chunk { } field access_control_list type weightedset { indexing: summary | attribute + rank: filter attribute: fast-search } field document_sets type weightedset { indexing: summary | attribute + rank: filter attribute: fast-search } }