Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

High Precision Audience Expansion via Extreme Classification in a Two-Sided Marketplace

Dillon Davis, Huiji Gao, Thomas Legrand, Juan Manuel Caicedo Carvajal, Malay Haldar, Kedar Bellare, Moutupsi Paul, Soumyadip Banerjee, Liwei He, Stephanie Moyerman, Sanjeev Katariya

TL;DR

The purpose of this paper is to demonstrate the methodology, challenges, and impact of rearchitecting search to retrieve from the subset of most bookable high precision rectangular map cells defined by dividing the world into 25M uniform cells.

Abstract

Airbnb search must balance a worldwide, highly varied supply of homes with guests whose location, amenity, style, and price expectations differ widely. Meeting those expectations hinges on an efficient retrieval stage that surfaces only the listings a guest might realistically book, before resource intensive ranking models are applied to determine the best results. Unlike many recommendation engines, our system faces a distinctive challenge, location retrieval, that sits upstream of ranking and determines which geographic areas are queried in order to filter inventory to a candidate set. The preexisting approach employs a deep bayesian bandit based system to predict a rectangular retrieval bounds area that can be used for filtering. The purpose of this paper is to demonstrate the methodology, challenges, and impact of rearchitecting search to retrieve from the subset of most bookable high precision rectangular map cells defined by dividing the world into 25M uniform cells.

High Precision Audience Expansion via Extreme Classification in a Two-Sided Marketplace

TL;DR

The purpose of this paper is to demonstrate the methodology, challenges, and impact of rearchitecting search to retrieve from the subset of most bookable high precision rectangular map cells defined by dividing the world into 25M uniform cells.

Abstract

Airbnb search must balance a worldwide, highly varied supply of homes with guests whose location, amenity, style, and price expectations differ widely. Meeting those expectations hinges on an efficient retrieval stage that surfaces only the listings a guest might realistically book, before resource intensive ranking models are applied to determine the best results. Unlike many recommendation engines, our system faces a distinctive challenge, location retrieval, that sits upstream of ranking and determines which geographic areas are queried in order to filter inventory to a candidate set. The preexisting approach employs a deep bayesian bandit based system to predict a rectangular retrieval bounds area that can be used for filtering. The purpose of this paper is to demonstrate the methodology, challenges, and impact of rearchitecting search to retrieve from the subset of most bookable high precision rectangular map cells defined by dividing the world into 25M uniform cells.
Paper Structure (21 sections, 6 equations, 6 figures, 1 table)

This paper contains 21 sections, 6 equations, 6 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Location retrieval in two-sided marketplaces.
  4. Spatial indexing and S2 geometry.
  5. Extreme multi-class classification.
  6. Joint retrieval and ranking.
  7. Approach
  8. Multiclass Classification
  9. Features
  10. Location Discretization
  11. Scalable Extreme Classification
  12. Architecture
  13. Retrieval Mechanism
  14. Experiments
  15. Dataset
  16. ...and 6 more sections

Figures (6)

  • Figure 1: Locations of Bookings from Cancun Searchers and Existing Retrieval Bounds
  • Figure 2: World Map split into S2 Cells
  • Figure 3: S2 Cell Model Output for Nashville, TN Search with Existing Retrieval Bounds
  • Figure 4: Europe S2 Cell Model Precision vs Recall
  • Figure 5: Americas S2 Cell Model Precision vs Recall
  • ...and 1 more figures