From c4d2a53f180c4eb6b8eb3cc8b56e0b037f84e67f Mon Sep 17 00:00:00 2001
From: Diverrez morgan <97022189+morgandiverrez@users.noreply.github.com>
Date: Fri, 26 Jul 2024 15:34:39 +0200
Subject: [PATCH] community: creation score_threshold in flashrank_rerank.py
 (#24016)

Description:
add a optional score relevance threshold for select only coherent
document, it's in complement of top_n

Discussion:
add relevance score threshold in flashrank_rerank document compressors
#24013

Dependencies:
 no dependencies

---------

Co-authored-by: Benjamin BERNARD <benjamin.bernard@openpathview.fr>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Chester Curme <chester.curme@gmail.com>
---
 .../document_compressors/flashrank_rerank.py  | 21 ++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)

diff --git a/libs/community/langchain_community/document_compressors/flashrank_rerank.py b/libs/community/langchain_community/document_compressors/flashrank_rerank.py
index fd66bee659d..64ea9a94191 100644
--- a/libs/community/langchain_community/document_compressors/flashrank_rerank.py
+++ b/libs/community/langchain_community/document_compressors/flashrank_rerank.py
@@ -26,8 +26,12 @@ class FlashrankRerank(BaseDocumentCompressor):
     """Flashrank client to use for compressing documents"""
     top_n: int = 3
     """Number of documents to return."""
+    score_threshold: float = 0.0
+    """Minimum relevance threshold to return."""
     model: Optional[str] = None
     """Model to use for reranking."""
+    prefix_metadata: str = ""
+    """Prefix for flashrank_rerank metadata keys"""
 
     class Config:
         """Configuration for this pydantic object."""
@@ -69,11 +73,14 @@ class FlashrankRerank(BaseDocumentCompressor):
         final_results = []
 
         for r in rerank_response:
-            metadata = r["meta"]
-            metadata["relevance_score"] = r["score"]
-            doc = Document(
-                page_content=r["text"],
-                metadata=metadata,
-            )
-            final_results.append(doc)
+            if r["score"] >= self.score_threshold:
+                doc = Document(
+                    page_content=r["text"],
+                    metadata={
+                        self.prefix_metadata + "id": r["id"],
+                        self.prefix_metadata + "relevance_score": r["score"],
+                        **r["meta"],
+                    },
+                )
+                final_results.append(doc)
         return final_results