1 minute read

CV Case Study: Subjective Scene Classification for Improved Search Experience

A leading hospitality marketplace partnered with Alegion to build a scalable data labeling pipeline for 30+ workflows ranging from competitor cross-listing to building type validation. Among those projects were ones focused on determining if a listing location was a primary residence, a secondary residence like a vacation home, or a dedicated rental property. There was also a need to classify what type of building a listing was in and if a private room was provided. The goal was to ensure that listing information that customers saw online was accurate and provided the best possible search experience.
 

This case study shows how Alegion improved accuracy from 60% to 98% through:

  • Platform configuration
  • Defining accuracy
  • Data exchange automation
  • Iterative testing process
  • Strategically applying learnings

The improvements in data quality improved model performance that drove the customer search experience. The additional insights gained have also allowed the client team to optimize internal resources, reducing operating costs and improving efficiency.

Download Now