harden semantic match sby tags

src/Intent/CatalogIntentLite.php

@@ -7,51 +7,52 @@ namespace App\Intent;
 use App\Tag\TagVectorSearchClient;
 use App\Tag\TagTypes;
 
+/**
+ * CatalogIntentLite
+ *
+ * Reiner Entity-Detector.
+ *
+ * Verantwortlich nur für:
+ * - Vector-Tag-Erkennung
+ * - Score-Gate
+ * - Ambiguity-Check
+ * - Sicherstellen, dass TagType = catalog_entity
+ *
+ * KEIN:
+ * - Listen-Signal
+ * - SalesIntent
+ * - Routing
+ */
 final class CatalogIntentLite
 {
-    private const LIST_SIGNALS = [
+    /**
-        'liste',
+     * Minimaler Similarity-Score.
-        'auflisten',
+     * Verhindert Rauschen.
-        'aufzaehl',
+     */
-        'aufzähl',
-        'übersicht',
-        'uebersicht',
-        'welche gibt es',
-        'welche sind',
-        'zeig mir alle',
-        'zeige mir alle',
-        'alle',
-    ];
-
-    // Realistischer Gate-Wert
     private const MIN_SCORE = 0.50;
 
-    // Ambiguity darf nicht zu aggressiv sein
+    /**
+     * Differenz zwischen Top1 und Top2,
+     * damit kein unsicherer Treffer akzeptiert wird.
+     */
     private const AMBIGUITY_DELTA = 0.01;
 
     public function __construct(
-        private readonly SalesIntentLite $salesIntentLite,
         private readonly TagVectorSearchClient $tagVectorClient,
     ) {}
 
+    /**
+     * Gibt das canonical Label der erkannten catalog_entity zurück
+     * oder null, wenn kein sauberer Treffer.
+     */
     public function detect(string $prompt): ?string
     {
-        $normalizedPrompt = mb_strtolower($prompt);
+        $prompt = trim($prompt);
-
+        if ($prompt === '') {
-        // 1) Listen-Signal prüfen
-        if (!$this->containsAny($normalizedPrompt, self::LIST_SIGNALS)) {
             return null;
         }
 
-        // 2) Nur DISCOVERY zulassen
+        // 1) Vector-Suche
-        $sales = $this->salesIntentLite->detect($prompt);
-        $intent = (string)($sales['intent'] ?? SalesIntentLite::DISCOVERY);
-
-        if ($intent !== SalesIntentLite::DISCOVERY) {
-            return null;
-        }
-
-        // 3) Vector-Search
         $hits = $this->tagVectorClient->search($prompt, 3);
 
         if ($hits === []) {
@@ -61,23 +62,26 @@ final class CatalogIntentLite
         $best = $hits[0];
         $bestScore = (float)($best['score'] ?? 0.0);
 
+        // 2) Score-Gate
         if ($bestScore < self::MIN_SCORE) {
             return null;
         }
 
-        // Ambiguity-Prüfung
+        // 3) Ambiguity-Check
         if (isset($hits[1])) {
             $secondScore = (float)($hits[1]['score'] ?? 0.0);
+
             if (abs($bestScore - $secondScore) < self::AMBIGUITY_DELTA) {
                 return null;
             }
         }
 
-        // Nur catalog_entity zulassen
+        // 4) Nur catalog_entity zulassen
         if (($best['tag_type'] ?? null) !== TagTypes::CATALOG_ENTITY) {
             return null;
         }
 
+        // 5) Canonical Label
         $label = trim((string)($best['label'] ?? ''));
 
         if ($label === '') {
@@ -86,15 +90,4 @@ final class CatalogIntentLite
 
         return mb_strtolower($label);
     }
-
-    private function containsAny(string $haystack, array $needles): bool
-    {
-        foreach ($needles as $needle) {
-            if ($needle !== '' && str_contains($haystack, $needle)) {
-                return true;
-            }
-        }
-
-        return false;
-    }
 }

src/Knowledge/Retrieval/NdjsonHybridRetriever.php

@@ -4,17 +4,15 @@ declare(strict_types=1);
 
 namespace App\Knowledge\Retrieval;
 
+use App\Catalog\EntityCatalogService;
 use App\Entity\ModelGenerationConfig;
 use App\Intent\CatalogIntentLite;
 use App\Intent\IntentLite;
 use App\Intent\SalesIntentLite;
-use App\Knowledge\QueryCleaner;
 use App\Repository\ModelGenerationConfigRepository;
+use App\Routing\IntentRouteResolver;
 use App\Tag\TagRoutingService;
 use App\Vector\VectorSearchClient;
-use App\Catalog\EntityCatalogService;
-use App\Knowledge\Retrieval\NdjsonChunkLookup;
-use App\Knowledge\Retrieval\RetrieverInterface;
 
 final class NdjsonHybridRetriever implements RetrieverInterface
 {
@@ -38,13 +36,13 @@ final class NdjsonHybridRetriever implements RetrieverInterface
         private readonly IntentLite                      $intentLite,
         private readonly SalesIntentLite                 $salesIntentLite,
         private readonly CatalogIntentLite               $catalogIntent,
+        private readonly IntentRouteResolver             $routeResolver,
         private readonly EntityCatalogService            $entityCatalogService
-    )
+    ) {
-    {
     }
 
     // =========================================================
-    // PRODUCTION (UNVERÄNDERTES VERHALTEN)
+    // PRODUCTION
     // =========================================================
 
     public function retrieve(string $prompt): array
@@ -60,17 +58,32 @@ final class NdjsonHybridRetriever implements RetrieverInterface
 
     public function retrieveInternal(string $prompt, ModelGenerationConfig $config): array
     {
-        // 🔵 ENTITY CATALOG EARLY EXIT (jetzt auch im Admin-Test aktiv)
+        // ------------------------------------------------------------
-        $entityTerm = $this->catalogIntent->detect($prompt);
+        // ROUTING-MATRIX (minimal, ohne Core zu zerlegen)
+        // ------------------------------------------------------------
 
-        if ($entityTerm !== null) {
+        // 1) Entity (semantisch über Tag-Vektor)
-            $catalogBlock = $this->entityCatalogService->listByTerm($entityTerm);
+        $entityLabel = $this->catalogIntent->detect($prompt);
+
+        // 2) Intent (regelbasiert)
+        $intent = (string)($this->salesIntentLite->detect($prompt)['intent'] ?? SalesIntentLite::DISCOVERY);
+
+        // 3) Route bestimmen (Intent + Entity)
+        $route = $this->routeResolver->resolve($intent, $entityLabel);
+
+        // 4) Early Exit nur für catalog_list
+        if ($route === IntentRouteResolver::ROUTE_CATALOG_LIST && $entityLabel !== null) {
+            $catalogBlock = $this->entityCatalogService->listByTerm($entityLabel);
 
             if ($catalogBlock !== null) {
                 return [$catalogBlock];
             }
         }
 
+        // ------------------------------------------------------------
+        // NORMALER CORE
+        // ------------------------------------------------------------
+
         $core = $this->runCore($prompt, $config, false);
 
         if ($core['ranked_chunk_ids'] === [] || $core['rows'] === []) {
@@ -95,7 +108,7 @@ final class NdjsonHybridRetriever implements RetrieverInterface
     }
 
     // =========================================================
-    // DEBUG (NEU, ABER NICHT IM PRODUKTIONS-PFAD)
+    // DEBUG (unverändert, kein Early-Exit damit Debug immer Core zeigt)
     // =========================================================
 
     /**

src/Knowledge/Retrieval/QueryCleaner.php

@@ -2,7 +2,9 @@
 
 declare(strict_types=1);
 
-namespace App\Knowledge;
+namespace App\Knowledge\Retrieval;
+
+use App\Knowledge\StopWords;
 
 final class QueryCleaner
 {

src/Routing/IntentRouteResolver.php

@@ -0,0 +1,70 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Routing;
+
+use App\Intent\SalesIntentLite;
+
+/**
+ * IntentRouteResolver
+ *
+ * Deterministische Routing-Matrix für:
+ *   Intent (SalesIntentLite)
+ *   +
+ *   erkannte Entity (CatalogIntentLite)
+ *
+ * Diese Klasse enthält KEINE Erkennungslogik.
+ * Sie entscheidet ausschließlich, welcher Modus gefahren wird.
+ *
+ * Erweiterbar über neue Intent-Konstanten.
+ */
+final class IntentRouteResolver
+{
+    public const ROUTE_NORMAL                = 'normal_rag';
+    public const ROUTE_CATALOG_LIST          = 'catalog_list';
+    public const ROUTE_ENTITY_PRICING        = 'entity_pricing';
+    public const ROUTE_ENTITY_COMPARISON     = 'entity_comparison';
+    public const ROUTE_ENTITY_IMPLEMENTATION = 'entity_implementation';
+    public const ROUTE_ENTITY_ROI            = 'entity_roi';
+
+    /**
+     * Routing-Entscheidung basierend auf Intent + Entity.
+     *
+     * @param string      $intent      Ergebnis aus SalesIntentLite
+     * @param string|null $entityLabel Ergebnis aus CatalogIntentLite
+     */
+    public function resolve(string $intent, ?string $entityLabel): string
+    {
+        // ------------------------------------------------------------
+        // 1) Keine Entity → normales RAG
+        // ------------------------------------------------------------
+        if ($entityLabel === null || $entityLabel === '') {
+            return self::ROUTE_NORMAL;
+        }
+
+        // ------------------------------------------------------------
+        // 2) Intent-basierte Entscheidung
+        // ------------------------------------------------------------
+        return match ($intent) {
+
+            SalesIntentLite::DISCOVERY
+            => self::ROUTE_CATALOG_LIST,
+
+            SalesIntentLite::PRICING
+            => self::ROUTE_ENTITY_PRICING,
+
+            SalesIntentLite::COMPARISON
+            => self::ROUTE_ENTITY_COMPARISON,
+
+            SalesIntentLite::IMPLEMENTATION
+            => self::ROUTE_ENTITY_IMPLEMENTATION,
+
+            SalesIntentLite::ROI
+            => self::ROUTE_ENTITY_ROI,
+
+            default
+            => self::ROUTE_NORMAL,
+        };
+    }
+}