Resumen
Per-record enriquecimiento flujo de trabajo for resolving B2B company names to canonical sitios web, handling rebrands y acquisitions via grafo de conocimiento alias datos. ~92-96% precision on limpiar names.
Desencadenador
Per CRM fila (batch o stream)
Programación
Per-record (batch o stream)
Pasos del flujo de trabajo
1
Scavio search per company nombre
Returns top organic + knowledge_graph entrada.
2
Pick candidate dominio
Prefer knowledge_graph.sitio web; fall back to top organic_results[0].enlace.
3
Scavio /extraer on candidate
Pull text de el candidate dominio home pagina.
4
Verify nombre match in extraido text
Lowercased substring o fuzzy match.
5
Compute confidence puntuacion
knowledge_graph hit + name-in-text + dominio no on un generic-host lista = high confidence.
6
Route low-confidence to human resena
Honest about el 4-8% edge cases.
Implementacion en Python
Python
import requests, os
H = {'x-api-key': os.environ['SCAVIO_API_KEY']}
def resolve(name):
s = requests.post('https://api.scavio.dev/api/v1/search', headers=H, json={'query': f'"{name}" official site'}).json()
kg = s.get('knowledge_graph', {})
candidate = kg.get('website') or (s.get('organic_results') or [{}])[0].get('link')
if not candidate:
return {'match': False}
page = requests.post('https://api.scavio.dev/api/v1/extract', headers=H, json={'url': candidate}).json()
text = (page.get('text') or '').lower()
return {'match': name.lower() in text, 'website': candidate}Implementacion en JavaScript
JavaScript
// Same shape in TS / Node — POST /api/v1/search then POST /api/v1/extract.Plataformas utilizadas
Búsqueda web con grafo de conocimiento, PAA y resúmenes de IA