SIFT: Search Intent-to-Filter Transformer for Multi-Task Personalized Filter Ranking at Airbnb

Key points

• Search filters help guests navigate vast catalogs in two-sided marketplaces like Airbnb, and recommending the right filters can meaningfully lift booking conversion.1
• Many such production filter-ranking systems, however, represent the guest through hand-engineered, pre-aggregated features generated by ETL pipelines.1
• SIFT replaces manual feature engineering with a unified guest representation that feeds multiple prediction tasks, including booking likelihood, filter engagement, and ordinal capacity thresholds (e.g., 2+ bedrooms) -- a general framework for filter ranking in two-sided marketplaces that accommodates both boolean and numeric-range filter types.1
• Demonstrating the system's extensibility, we rapidly integrated a novel hotel-intent filter using the same shared representation, driving a +3.8% lift in uncancelled hotel bookings and a +0.76% lift in overall marketplace bookings.1

Sources (1)

• [1]SIFT: Search Intent-to-Filter Transformer for Multi-Task Personalized Filter Ranking at Airbnb
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 06:05 AM

  We present SIFT (Search Intent-to-Filter Transformer), a ranking model built on transformers that learns guest preferences directly from raw behavioral sequences.

  Search filters help guests navigate vast catalogs in two-sided marketplaces like Airbnb, and recommending the right filters can meaningfully lift booking conversion.

Extractive summary: sentences quoted from the sources.