From 4761648836d5619e90a5dd06d4007a2afbbf72d4 Mon Sep 17 00:00:00 2001
From: team 1 <team1@mitho-media.de>
Date: Fri, 27 Feb 2026 12:03:45 +0100
Subject: [PATCH] new system rebuild command harden IngestService

---
 .../Ingest/KnowledgeIngestService.php         | 26 +++++++++----------
 .../Retrieval/NdjsonHybridRetriever.php       |  4 +--
 2 files changed, 15 insertions(+), 15 deletions(-)

diff --git a/src/Knowledge/Ingest/KnowledgeIngestService.php b/src/Knowledge/Ingest/KnowledgeIngestService.php
index adf299f..acf3b14 100644
--- a/src/Knowledge/Ingest/KnowledgeIngestService.php
+++ b/src/Knowledge/Ingest/KnowledgeIngestService.php
@@ -10,10 +10,11 @@ use App\Repository\DocumentVersionRepository;
 final class KnowledgeIngestService
 {
     public function __construct(
-        private DocumentLoader $loader,
-        private SimpleChunker  $chunker,
+        private DocumentLoader            $loader,
+        private SimpleChunker             $chunker,
         private DocumentVersionRepository $versionRepo,
-    ) {
+    )
+    {
     }
 
     /**
@@ -31,17 +32,16 @@ final class KnowledgeIngestService
         $doc = $version->getDocument();
 
         $documentId = $doc->getId()->toRfc4122();
-        $versionId  = $version->getId()->toRfc4122();
+        $versionId = $version->getId()->toRfc4122();
 
-        $title = trim((string) $doc->getTitle());
+        $title = trim((string)$doc->getTitle());
 
         $index = 0;
 
         foreach ($chunks as $chunkText) {
 
-            // Titel optional weiterhin prefixen (wenn du das behalten willst)
             if ($title !== '' && !str_starts_with($chunkText, $title)) {
-                $chunkText = $title . "\n\n" . $chunkText;
+                $chunkText = "# Produkt Titel: " . $title . "\n\n --- " . $chunkText;
             }
 
             $chunkText = trim($chunkText);
@@ -56,13 +56,13 @@ final class KnowledgeIngestService
             );
 
             yield [
-                'chunk_id'    => $chunkId,
+                'chunk_id' => $chunkId,
                 'document_id' => $documentId,
-                'version_id'  => $versionId,
+                'version_id' => $versionId,
                 'chunk_index' => $index++,
-                'text'        => $chunkText,
-                'checksum'    => sha1($chunkText),
-                'metadata'    => $this->buildMetadata($version),
+                'text' => $chunkText,
+                'checksum' => sha1($chunkText),
+                'metadata' => $this->buildMetadata($version),
             ];
         }
     }
@@ -114,7 +114,7 @@ final class KnowledgeIngestService
             'version_number' => method_exists($version, 'getVersionNumber')
                 ? $version->getVersionNumber()
                 : null,
-            'file_path'      => $version->getFilePath(),
+            'file_path' => $version->getFilePath(),
         ], static fn($v) => $v !== null && $v !== '');
     }
 }
\ No newline at end of file
diff --git a/src/Knowledge/Retrieval/NdjsonHybridRetriever.php b/src/Knowledge/Retrieval/NdjsonHybridRetriever.php
index 64d3244..43e616f 100644
--- a/src/Knowledge/Retrieval/NdjsonHybridRetriever.php
+++ b/src/Knowledge/Retrieval/NdjsonHybridRetriever.php
@@ -14,7 +14,7 @@ use App\Vector\VectorSearchClient;
 
 final class NdjsonHybridRetriever implements RetrieverInterface
 {
-    private const VECTOR_SCORE_THRESHOLD = 0.4;
+    private const VECTOR_SCORE_THRESHOLD = 0.75;
 
     private const HARD_MAX_CHUNKS = 200;
     private const HARD_MAX_VECTORK = 200;
@@ -23,7 +23,7 @@ final class NdjsonHybridRetriever implements RetrieverInterface
      * Tags dürfen nur ein kleiner Bonus sein (kein Gate/Filter).
      * Enterprise Default: klein halten, sonst dominieren Tags wieder.
      */
-    private const TAG_SCORE_BONUS = 0.08;
+    private const TAG_SCORE_BONUS = 0.25;
 
     public function __construct(
         private readonly NdjsonChunkLookup               $lookup,