Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Learning to Rank for Maps at Airbnb

Malay Haldar, Hongwei Zhang, Kedar Bellare, Sherry Chen, Soumyadip Banerjee, Xiaotang Wang, Mustafa Abdool, Huiji Gao, Pavan Tapadia, Liwei He, Sanjeev Katariya

TL;DR

This work reveals that ranking strategies effective for list-based Airbnb search do not translate to map-based interfaces due to fundamentally different user attention patterns. It introduces a map-specific attention framework, notably the Less Is More principle with the Bookability Filter and a robust anchor mechanism, plus tiered pins and a 2D CTR model to optimize map pins and center positioning. Through offline simulations and large-scale online A/B tests, the authors demonstrate meaningful gains in bookings and reduced cognitive load, validating a dedicated map-focused learning-to-rank approach. The results have practical impact by improving user experience and conversion on map searches, while also charting a path for future map-specific benchmarks and loss-function design in ranking research.

Abstract

As a two-sided marketplace, Airbnb brings together hosts who own listings for rent with prospective guests from around the globe. Results from a guest's search for listings are displayed primarily through two interfaces: (1) as a list of rectangular cards that contain on them the listing image, price, rating, and other details, referred to as list-results (2) as oval pins on a map showing the listing price, called map-results. Both these interfaces, since their inception, have used the same ranking algorithm that orders listings by their booking probabilities and selects the top listings for display. But some of the basic assumptions underlying ranking, built for a world where search results are presented as lists, simply break down for maps. This paper describes how we rebuilt ranking for maps by revising the mathematical foundations of how users interact with search results. Our iterative and experiment-driven approach led us through a path full of twists and turns, ending in a unified theory for the two interfaces. Our journey shows how assumptions taken for granted when designing machine learning algorithms may not apply equally across all user interfaces, and how they can be adapted. The net impact was one of the largest improvements in user experience for Airbnb which we discuss as a series of experimental validations.

Learning to Rank for Maps at Airbnb

TL;DR

This work reveals that ranking strategies effective for list-based Airbnb search do not translate to map-based interfaces due to fundamentally different user attention patterns. It introduces a map-specific attention framework, notably the Less Is More principle with the Bookability Filter and a robust anchor mechanism, plus tiered pins and a 2D CTR model to optimize map pins and center positioning. Through offline simulations and large-scale online A/B tests, the authors demonstrate meaningful gains in bookings and reduced cognitive load, validating a dedicated map-focused learning-to-rank approach. The results have practical impact by improving user experience and conversion on map searches, while also charting a path for future map-specific benchmarks and loss-function design in ranking research.

Abstract

As a two-sided marketplace, Airbnb brings together hosts who own listings for rent with prospective guests from around the globe. Results from a guest's search for listings are displayed primarily through two interfaces: (1) as a list of rectangular cards that contain on them the listing image, price, rating, and other details, referred to as list-results (2) as oval pins on a map showing the listing price, called map-results. Both these interfaces, since their inception, have used the same ranking algorithm that orders listings by their booking probabilities and selects the top listings for display. But some of the basic assumptions underlying ranking, built for a world where search results are presented as lists, simply break down for maps. This paper describes how we rebuilt ranking for maps by revising the mathematical foundations of how users interact with search results. Our iterative and experiment-driven approach led us through a path full of twists and turns, ending in a unified theory for the two interfaces. Our journey shows how assumptions taken for granted when designing machine learning algorithms may not apply equally across all user interfaces, and how they can be adapted. The net impact was one of the largest improvements in user experience for Airbnb which we discuss as a series of experimental validations.
Paper Structure (14 sections, 6 equations, 13 figures, 3 tables, 2 algorithms)

This paper contains 14 sections, 6 equations, 13 figures, 3 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Map $\neq$ List
  3. Experimental Results
  4. Less Is More
  5. Experimental Results
  6. Making It Robust
  7. Experimental Results
  8. Focus vs. Urgency?
  9. Experimental Results
  10. Tiered User Attention
  11. Experimental Results
  12. How Attention Flows On Maps
  13. Experimental Results
  14. Insights And Future Work

Figures (13)

  • Figure 1: Search box with destination, checkin/checkout dates and guest count as inputs.
  • Figure 2: Search results as map pins.
  • Figure 3: Search results as list of cards.
  • Figure 4: X-axis: Search rank for list-results. Y-axis: Normalized click-through rates.
  • Figure 5: X-axis: Search rank for list-results and map-results. Y-axis: Normalized click-through rates.
  • ...and 8 more figures