From 92f543c48f2230e20a8eb9e0b554ce7ff7eac189 Mon Sep 17 00:00:00 2001
From: Jacob <cj0336j@gmail.com>
Date: Sun, 6 Sep 2026 13:18:09 +0900
Subject: [PATCH] =?UTF-8?q?feat(embedding):=20=EC=B2=AD=ED=81=AC=20?=
 =?UTF-8?q?=EB=B6=84=ED=95=A0=20=EB=8B=A8=EC=9C=84=20450=20=ED=86=A0?=
 =?UTF-8?q?=ED=81=B0=EC=9C=BC=EB=A1=9C=20=EC=B5=9C=EC=A0=81=ED=99=94?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 pipelines/embedding/document_parser.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/pipelines/embedding/document_parser.py b/pipelines/embedding/document_parser.py
index 47118b0..d6a32a4 100644
--- a/pipelines/embedding/document_parser.py
+++ b/pipelines/embedding/document_parser.py
@@ -189,8 +189,8 @@ class DocumentParser:
 
     def __init__(self) -> None:
         self._chunker = FinancialChunker(
-            chunk_size=750,  # 한국어 금융 문맥 보존용 최적 크기
-            chunk_overlap=150,
+            chunk_size=450,  # 한국어 금융 문맥 보존 및 정밀 검색용 최적 크기 (주식과 세금 단락 기준 통일)
+            chunk_overlap=100,
         )
 
     def parse_file(self, path: Path) -> list[Passage]: