fix shop research intent

RETRIEX_SHOP_ADVISORY_INTENT_FIX_README.md

@@ -0,0 +1,64 @@
+# RetrieX shop advisory intent fix
+
+## Problem
+
+The prompt
+
+```text
+welcher pockettester ist für Redox messung gut
+```
+
+was classified as a purely technical RAG question because the factual guard matched
+question wording such as `welcher` plus technical wording such as `Messung`.
+As a result no Shopware search was executed. The later follow-up
+
+```text
+suche im shop
+```
+
+then had to recover a shop query from history/client context, which remained fragile.
+
+## Fix
+
+This patch keeps the factual RAG guard for technical questions, but lets clear product
+advisory prompts pass into commerce detection when both conditions are true:
+
+1. The prompt contains an advisory term such as `gut`, `geeignet`, `empfehl...`.
+2. The prompt contains a product/device class signal such as `pockettester`,
+   `messgerät`, `handmessgerät`, etc.
+
+This means:
+
+- `Was ist der niedrigste Grenzwert ... Testomat ... Wasserhärte?` stays RAG-only.
+- `mit welchem indikator wird der wert gemessen` stays RAG-only.
+- `was kostet der indikator` stays commerce/product search.
+- `welcher pockettester ist für Redox messung gut` now triggers commerce search directly.
+
+## Changed files
+
+```text
+src/Intent/CommerceIntentLite.php
+src/Config/CommerceIntentConfig.php
+config/retriex/intent.yaml
+config/retriex/agent.yaml
+```
+
+## Validation performed
+
+```bash
+php -n -l src/Intent/CommerceIntentLite.php
+php -n -l src/Config/CommerceIntentConfig.php
+```
+
+Additional local intent check:
+
+```text
+Was ist der niedrigste Grenzwert ... => none
+mit welchem indikator ... => none
+was kostet der indikator => product_search
+welcher pockettester ist für Redox messung gut => product_search
+suche im shop => product_search
+suche im shop nach redox messung => product_search
+welcher Grenzwert kann mit dem Testomat 808 gemessen werden => none
+welches Messgerät ist für Redox Messung geeignet => product_search
+```

config/retriex/agent.yaml

@@ -61,7 +61,7 @@ parameters:
         - '- Separate terms using spaces only.'
         - '- If a relevant product name is present, it must be placed at the beginning of the final search query.'
         - '- Try to always identify all products mentioned in the user input text, even in long prompts.'
-        - '- Look for terms such as Testomat, Horiba, Tritromat, or words like indicator/Indikator.'
+        - '- Look for terms such as Testomat, Horiba, Tritromat, Pockettester, Redox, ORP, or words like indicator/Indikator.'
         - '- If the current user input is vague or referential, use the recent conversation context only as support.'
         - '- Do not output words that only describe conversation flow, such as "same", "again", "also", or "like above".'
       conversation_context_rules:

config/retriex/intent.yaml

@@ -21,6 +21,10 @@ parameters:
       - analysegeraet
       - messgerät
       - messgeraet
+      - pockettester
+      - pocket tester
+      - handmessgerät
+      - handmessgeraet
       - analysator
       - analyzer
       - puffer
@@ -40,6 +44,7 @@ parameters:
       - eignet
       - besser
       - besten
+      - gut
       - gut für
       - gut fuer
       - passend für

public/assets/styles/base.css

@@ -393,7 +393,7 @@ span.think {
     background-position: 220% 0;
     -webkit-background-clip: text;
     background-clip: text;
-    animation: thinkScan 2.2s linear infinite;
+    animation: thinkScan 1.4s linear infinite;
 }
 
 @keyframes thinkScan {

src/Config/CommerceIntentConfig.php

@@ -25,6 +25,10 @@ final class CommerceIntentConfig
         'analysegeraet',
         'messgerät',
         'messgeraet',
+        'pockettester',
+        'pocket tester',
+        'handmessgerät',
+        'handmessgeraet',
         'analysator',
         'analyzer',
         'puffer',
@@ -46,6 +50,7 @@ final class CommerceIntentConfig
         'eignet',
         'besser',
         'besten',
+        'gut',
         'gut für',
         'gut fuer',
         'passend für',

src/Intent/CommerceIntentLite.php

@@ -40,7 +40,11 @@ final class CommerceIntentLite
             );
         }
 
-        if ($this->isTechnicalFactualKnowledgeQuery($prompt) && !$this->hasExplicitCommerceIntent($prompt)) {
+        if (
+            $this->isTechnicalFactualKnowledgeQuery($prompt)
+            && !$this->hasExplicitCommerceIntent($prompt)
+            && !$this->hasAdvisoryProductSelectionIntent($prompt)
+        ) {
             return $this->buildDetectionResult(
                 intent: self::NONE,
                 score: 0,
@@ -95,6 +99,81 @@ final class CommerceIntentLite
         return $this->matchesAnyPattern($prompt, $this->config->getExplicitCommerceIntentPatterns());
     }
 
+    /**
+     * Product advisory prompts can contain technical words such as "Messung".
+     *
+     * They must not be swallowed by the factual RAG guard when the user is
+     * clearly asking for a suitable shop product or device class.
+     */
+    private function hasAdvisoryProductSelectionIntent(string $prompt): bool
+    {
+        if (!$this->containsConfiguredAdvisorySignal($prompt)) {
+            return false;
+        }
+
+        if (preg_match($this->config->getModelLikeProductPattern(), $prompt) === 1) {
+            return true;
+        }
+
+        foreach ($this->config->getStrongSignalsList() as $signal) {
+            $signal = mb_strtolower(trim($signal));
+
+            if ($signal === '' || $this->isNonProductCommerceSignal($signal)) {
+                continue;
+            }
+
+            if (str_contains($prompt, $signal)) {
+                return true;
+            }
+        }
+
+        return false;
+    }
+
+    private function containsConfiguredAdvisorySignal(string $prompt): bool
+    {
+        foreach ($this->config->getAdvisorySignals() as $signal) {
+            if ($this->containsConfiguredPhraseOrToken($prompt, $signal)) {
+                return true;
+            }
+        }
+
+        return false;
+    }
+
+    private function containsConfiguredPhraseOrToken(string $prompt, string $signal): bool
+    {
+        $signal = mb_strtolower(trim($signal));
+
+        if ($signal === '') {
+            return false;
+        }
+
+        if (preg_match('/\s/u', $signal) === 1) {
+            return str_contains($prompt, $signal);
+        }
+
+        $pattern = '/(?<![\p{L}\p{N}])' . preg_quote($signal, '/') . '(?![\p{L}\p{N}])/u';
+
+        return preg_match($pattern, $prompt) === 1;
+    }
+
+    private function isNonProductCommerceSignal(string $signal): bool
+    {
+        return in_array($signal, [
+            'shop',
+            'alle',
+            'kunde',
+            'online',
+            'kaufen',
+            'kostet',
+            'suche',
+            'such',
+            'finde',
+            'finden',
+        ], true);
+    }
+
     /**
      * Detects factual technical knowledge questions that must stay in RAG retrieval.
      *
@@ -255,7 +334,7 @@ final class CommerceIntentLite
     private function applyAdvisorySignals(string $prompt, int $score, array $signals): array
     {
         foreach ($this->config->getAdvisorySignals() as $signal) {
-            if (str_contains($prompt, mb_strtolower($signal))) {
+            if ($this->containsConfiguredPhraseOrToken($prompt, $signal)) {
                 $score += $this->config->getAdvisorySignalScore();
                 $signals[] = $this->config->getAdvisorySignalPrefix() . $signal;
             }