feat(classifier): add multilingual ECP inherence classifier POC

2026-08-20 00:51:02 -03:00
parent d371b81aa4
commit 67cc40f91a
175 changed files with 30399 additions and 703 deletions
@@ -0,0 +1,2 @@
+# Batteriefabrik in Europa
+Northvolt steigert die Produktion von Batteriezellen für führende Automobilhersteller auf dem europäischen Markt.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "CONTEXTUAL_INHERENT",
+  "expected_is_inherent": true,
+  "expected_language": "de",
+  "min_confidence": 0.70
+}
@@ -0,0 +1,2 @@
+# Investitionsoffensive in Wolfsburg
+Volkswagen investiert Milliarden in die Fahrzeugproduktion und neue Plattformen für moderne Elektrofahrzeuge.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "DIRECT_INHERENT",
+  "expected_is_inherent": true,
+  "expected_language": "de",
+  "min_confidence": 0.85
+}
@@ -0,0 +1,18 @@
+{
+  "target_entity_id": "ent_volkswagen_de",
+  "target_name": "Volkswagen",
+  "aliases": ["Volkswagen AG", "VW", "Volkswagen Group"],
+  "domain": "Automobilindustrie",
+  "anchors": ["Elektrofahrzeuge", "Batteriezellen", "Fahrzeugproduktion", "Automobilhersteller", "Elektromobilität"],
+  "negative_anchors": ["Modellautosammlung", "Spielzeugautos"],
+  "graph_version": "1.0.0",
+  "related_entities": [
+    {
+      "entity_id": "ent_northvolt_de",
+      "name": "Northvolt",
+      "relation_type": "SUPPLIER_OF",
+      "weight": 0.85,
+      "scope": "battery_supply"
+    }
+  ]
+}
@@ -0,0 +1,2 @@
+# Backrezept für Apfelstrudel
+Den Teig dünn ausrollen, mit Rosinen und Zimt bestreuen und bei mittlerer Hitze im Ofen goldgelb backen.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "NOT_RELATED",
+  "expected_is_inherent": false,
+  "expected_language": "de",
+  "min_confidence": 0.85
+}
@@ -0,0 +1,2 @@
+# Reisebericht aus Niedersachsen
+Beim Spaziergang durch die Fußgängerzone sahen wir in der Ferne ein Werbeschild von Volkswagen neben dem Parkhaus.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "TANGENTIAL",
+  "expected_is_inherent": false,
+  "expected_language": "de",
+  "min_confidence": 0.30
+}
@@ -0,0 +1,2 @@
+# Electronics Assembly Expansion
+Foxconn is expanding its major manufacturing plants to boost production capacity for consumer electronics devices.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "CONTEXTUAL_INHERENT",
+  "expected_is_inherent": true,
+  "expected_language": "en",
+  "min_confidence": 0.70
+}
@@ -0,0 +1,3 @@
+# Apple Unveils Next-Generation iPhone
+
+Apple announced its flagship iPhone today, featuring an advanced custom silicon chip and upgraded camera hardware.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "DIRECT_INHERENT",
+  "expected_is_inherent": true,
+  "expected_language": "en",
+  "min_confidence": 0.85
+}
@@ -0,0 +1,22 @@
+{
+  "target_entity_id": "ent_apple_en",
+  "target_name": "Apple",
+  "aliases": ["Apple Inc.", "Apple"],
+  "domain": "Technology & Consumer Electronics",
+  "anchors": ["iPhone", "MacBook", "iOS", "silicon", "hardware", "smartphone", "consumer electronics", "manufacturing"],
+  "negative_anchors": [
+    "apple pie",
+    "apple pie recipe",
+    "orchard harvest"
+  ],
+  "graph_version": "1.0.0",
+  "related_entities": [
+    {
+      "entity_id": "ent_foxconn_en",
+      "name": "Foxconn",
+      "relation_type": "MANUFACTURER_FOR",
+      "weight": 0.85,
+      "scope": "manufacturing"
+    }
+  ]
+}
@@ -0,0 +1,2 @@
+# Classic Apple Pie Baking Guide
+Mix sliced cinnamon apples with brown sugar and butter before placing inside the homemade flaky apple pie crust.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "NOT_RELATED",
+  "expected_is_inherent": false,
+  "expected_language": "en",
+  "min_confidence": 0.85
+}
@@ -0,0 +1,2 @@
+# Morning City Walk
+While walking through the park downtown, we noticed an Apple poster near the transit stop before heading into the coffee shop.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "TANGENTIAL",
+  "expected_is_inherent": false,
+  "expected_language": "en",
+  "min_confidence": 0.30
+}
@@ -0,0 +1,2 @@
+# Nuevas Soluciones de Pagos Electrónicos
+La compañía PagoNxt implementó un nuevo sistema de liquidación de pagos y financiación para comercios internacionales.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "CONTEXTUAL_INHERENT",
+  "expected_is_inherent": true,
+  "expected_language": "es",
+  "min_confidence": 0.70
+}
@@ -0,0 +1,2 @@
+# Expansión de Servicios Financieros en España
+El Banco Santander anunció hoy un aumento en la concesión de créditos e hipotecas a través de su plataforma de banca digital.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "DIRECT_INHERENT",
+  "expected_is_inherent": true,
+  "expected_language": "es",
+  "min_confidence": 0.85
+}
@@ -0,0 +1,18 @@
+{
+  "target_entity_id": "ent_santander_es",
+  "target_name": "Banco Santander",
+  "aliases": ["Santander", "Grupo Santander", "Banco Santander S.A."],
+  "domain": "Banca y Finanzas",
+  "anchors": ["créditos", "sucursales", "cuentas bancarias", "hipotecas", "financiación", "banca digital"],
+  "negative_anchors": ["playa de santander", "bahía de santander turismo"],
+  "graph_version": "1.0.0",
+  "related_entities": [
+    {
+      "entity_id": "ent_pagonxt_es",
+      "name": "PagoNxt",
+      "relation_type": "SUBSIDIARY_OF",
+      "weight": 0.85,
+      "scope": "payments"
+    }
+  ]
+}
@@ -0,0 +1,2 @@
+# Turismo en la Costa Norte
+La hermosa bahía de santander turismo ofrece paseos marítimos y playas con vistas espectaculares del mar Cantábrico.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "NOT_RELATED",
+  "expected_is_inherent": false,
+  "expected_language": "es",
+  "min_confidence": 0.85
+}
@@ -0,0 +1,2 @@
+# Guía Cultural de la Ciudad
+Caminando cerca del ayuntamiento pasamos frente a un edificio del Santander en una tarde soleada de primavera.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "TANGENTIAL",
+  "expected_is_inherent": false,
+  "expected_language": "es",
+  "min_confidence": 0.30
+}
@@ -0,0 +1,2 @@
+# Batteries Industrielles de Pointe
+L'entreprise Saft a inauguré une ligne de fabrication de batteries avancées destinées au stockage d'énergie et aux transports.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "CONTEXTUAL_INHERENT",
+  "expected_is_inherent": true,
+  "expected_language": "fr",
+  "min_confidence": 0.70
+}
@@ -0,0 +1,2 @@
+# Développement Énergétique en France
+TotalEnergies a annoncé de nouveaux investissements massifs dans les énergies renouvelables et le raffinage durable.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "DIRECT_INHERENT",
+  "expected_is_inherent": true,
+  "expected_language": "fr",
+  "min_confidence": 0.85
+}
@@ -0,0 +1,18 @@
+{
+  "target_entity_id": "ent_totalenergies_fr",
+  "target_name": "TotalEnergies",
+  "aliases": ["TotalEnergies SE", "Total"],
+  "domain": "Énergie et Pétrole",
+  "anchors": ["raffinage", "énergies renouvelables", "carburants", "gaz naturel", "production pétrolière", "stockage d'énergie", "batteries"],
+  "negative_anchors": ["total look mode", "somme totale facture"],
+  "graph_version": "1.0.0",
+  "related_entities": [
+    {
+      "entity_id": "ent_saft_fr",
+      "name": "Saft",
+      "relation_type": "SUBSIDIARY_OF",
+      "weight": 0.85,
+      "scope": "battery_technology"
+    }
+  ]
+}
@@ -0,0 +1,2 @@
+# Recette Traditionnelle de la Quiche Lorraine
+Mélanger la crème fraîche avec les œufs battus, ajouter les lardons dorés et verser sur la pâte brisée avant de cuire au four.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "NOT_RELATED",
+  "expected_is_inherent": false,
+  "expected_language": "fr",
+  "min_confidence": 0.85
+}
@@ -0,0 +1,2 @@
+# Promenade dans Paris
+En marchant le long du boulevard Haussmann, nous avons aperçu un panneau publicitaire de TotalEnergies près de la station de métro.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "TANGENTIAL",
+  "expected_is_inherent": false,
+  "expected_language": "fr",
+  "min_confidence": 0.30
+}
@@ -0,0 +1,2 @@
+# Innovazione negli Impianti Frenanti
+Brembo ha sviluppato una nuova generazione di freni carboceramici ad alte prestazioni per le supercar sportive.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "CONTEXTUAL_INHERENT",
+  "expected_is_inherent": true,
+  "expected_language": "it",
+  "min_confidence": 0.70
+}
@@ -0,0 +1,2 @@
+# Nuova Monoposto a Maranello
+La Ferrari ha presentato la nuova vettura di Formula 1 dotata di motori ibridi potenziati e aerodinamica avanzata.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "DIRECT_INHERENT",
+  "expected_is_inherent": true,
+  "expected_language": "it",
+  "min_confidence": 0.85
+}
@@ -0,0 +1,18 @@
+{
+  "target_entity_id": "ent_ferrari_it",
+  "target_name": "Ferrari",
+  "aliases": ["Scuderia Ferrari", "Ferrari N.V."],
+  "domain": "Supercar e Motorsport",
+  "anchors": ["Maranello", "motori", "Formula 1", "supercar", "aerodinamica", "velocità"],
+  "negative_anchors": ["Ferrari Trento vino spumante"],
+  "graph_version": "1.0.0",
+  "related_entities": [
+    {
+      "entity_id": "ent_brembo_it",
+      "name": "Brembo",
+      "relation_type": "SUPPLIER_OF",
+      "weight": 0.85,
+      "scope": "braking_systems"
+    }
+  ]
+}
@@ -0,0 +1,2 @@
+# Ricetta Tradizionale del Risotto
+Tostare il riso Carnaroli con burro e cipolla, sfumare con brodo caldo e mantecare con formaggio Parmigiano Reggiano.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "NOT_RELATED",
+  "expected_is_inherent": false,
+  "expected_language": "it",
+  "min_confidence": 0.85
+}
@@ -0,0 +1,2 @@
+# Passeggiata Pomeridiana a Modena
+Camminando nel centro storico di Modena abbiamo notato una maglietta della Ferrari esposta nella vetrina di un negozio.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "TANGENTIAL",
+  "expected_is_inherent": false,
+  "expected_language": "it",
+  "min_confidence": 0.30
+}
@@ -0,0 +1,2 @@
+# Logística de Derivados
+A Transpetro modernizou os dutos e navios petroleiros para transporte de combustíveis pelo país.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "CONTEXTUAL_INHERENT",
+  "expected_is_inherent": true,
+  "expected_language": "pt",
+  "min_confidence": 0.70
+}
@@ -0,0 +1,2 @@
+# Produção de Petróleo no Brasil
+A Petrobras registrou um aumento na extração de petróleo no pré-sal com novas tecnologias offshore.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "DIRECT_INHERENT",
+  "expected_is_inherent": true,
+  "expected_language": "pt",
+  "min_confidence": 0.85
+}
@@ -0,0 +1,18 @@
+{
+  "target_entity_id": "ent_petrobras_pt",
+  "target_name": "Petrobras",
+  "aliases": ["Petróleo Brasileiro S.A.", "Petrobras"],
+  "domain": "Energia e Petróleo",
+  "anchors": ["pré-sal", "refinaria", "combustíveis", "petróleo"],
+  "negative_anchors": ["posto clandestino"],
+  "graph_version": "1.0.0",
+  "related_entities": [
+    {
+      "entity_id": "ent_transpetro_pt",
+      "name": "Transpetro",
+      "relation_type": "SUBSIDIARY_OF",
+      "weight": 0.9,
+      "scope": "logistics"
+    }
+  ]
+}
@@ -0,0 +1,2 @@
+# Dicas de Jardinagem
+Cultivar orquídeas requer rega moderada e ambiente com luz solar indireta para florescer com saúde.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "NOT_RELATED",
+  "expected_is_inherent": false,
+  "expected_language": "pt",
+  "min_confidence": 0.85
+}
@@ -0,0 +1,2 @@
+# Turismo no Rio de Janeiro
+Durante a caminhada pela orla da praia, avistamos ao longe uma placa da Petrobras perto da avenida movimentada.
@@ -0,0 +1,6 @@
+{
+  "expected_decision": "TANGENTIAL",
+  "expected_is_inherent": false,
+  "expected_language": "pt",
+  "min_confidence": 0.30
+}