POI & Media Data Pipeline

What is it?

The Data Pipeline collects, validates and enriches destination data in three controlled phases, in accordance with the Medallion principle from data engineering. Raw data from multiple sources is first checked and cleaned. This is followed by enrichment using AI and publication in every language that the destination has activated.

Specifications

• A range of reputable sources side by side: Google Places, OpenStreetMap, official websites and tourist portals. No single source is ever considered the definitive truth.
• Three quality control stages: initial collection of locations, diaries, reviews, weather and media; verification and data cleansing across more than eighty data fields per location; enrichment using AI (Mistral, France) and translation with built-in quality control.
• A dynamic database with no limits: thousands of records per destination today, designed to scale up to hundreds of thousands.
• Every activated language is fully covered and automatically monitored with a weekly coverage check.
• Freshness is monitored for each record. Change detection automatically flags closures and falling ratings.
• The update frequency can be set for each holding, ranging from hourly to quarterly, in line with strategy and budget.
• Review analysis based on tens of thousands of reviews. The website only features up-to-date information, never more than two years old.
• Calendar integration: events and iCal feeds are automatically synchronised.