Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The Dating Heuristic: A Provably Strong Matching Algorithm for Dating Platforms

Ignacio Rios, Alfredo Torrico

TL;DR

The Integral Dating Heuristic (DH-int), providing formal performance guarantees: DH-int achieves a uniform $1-1/e$ guarantee across all platform designs under reasonable assumptions, and approaches the theoretical upper bound across multiple platform designs and variants of the history effect.

Abstract

Motivated by online dating platforms, we study the problem of selecting which subset of profiles to display to each user in each period. Users observe the profiles set by the platform, decide which of them to like, and a match occurs if and only if two users mutually like each other, potentially across different periods. The platform aims to maximize the expected number of matches produced over the entire time horizon, and users' behavior -- captured by their like probabilities -- may depend on their history. We develop a general theoretical model that captures the dynamic, two-sided nature of the problem and the influence of users' past experiences on their future behavior. We focus on one-lookahead policies and propose the Integral Dating Heuristic (DH-int), providing formal performance guarantees: DH-int achieves a uniform $1-1/e$ guarantee across all platform designs under reasonable assumptions. Our empirical analysis, using proprietary data from a major U.S.-based dating app, confirms that DH-int consistently outperforms other benchmarks such as Greedy, Perfect Matching and DH, and approaches the theoretical upper bound across multiple platform designs and variants of the history effect. The superior performance of DH-int is driven primarily by its careful balancing of initial and follow-up interactions, which accounts for the two-sided nature of the market.

The Dating Heuristic: A Provably Strong Matching Algorithm for Dating Platforms

TL;DR

The Integral Dating Heuristic (DH-int), providing formal performance guarantees: DH-int achieves a uniform guarantee across all platform designs under reasonable assumptions, and approaches the theoretical upper bound across multiple platform designs and variants of the history effect.

Abstract

Motivated by online dating platforms, we study the problem of selecting which subset of profiles to display to each user in each period. Users observe the profiles set by the platform, decide which of them to like, and a match occurs if and only if two users mutually like each other, potentially across different periods. The platform aims to maximize the expected number of matches produced over the entire time horizon, and users' behavior -- captured by their like probabilities -- may depend on their history. We develop a general theoretical model that captures the dynamic, two-sided nature of the problem and the influence of users' past experiences on their future behavior. We focus on one-lookahead policies and propose the Integral Dating Heuristic (DH-int), providing formal performance guarantees: DH-int achieves a uniform guarantee across all platform designs under reasonable assumptions. Our empirical analysis, using proprietary data from a major U.S.-based dating app, confirms that DH-int consistently outperforms other benchmarks such as Greedy, Perfect Matching and DH, and approaches the theoretical upper bound across multiple platform designs and variants of the history effect. The superior performance of DH-int is driven primarily by its careful balancing of initial and follow-up interactions, which accounts for the two-sided nature of the market.
Paper Structure (48 sections, 27 theorems, 92 equations, 5 figures, 4 tables, 6 algorithms)

This paper contains 48 sections, 27 theorems, 92 equations, 5 figures, 4 tables, 6 algorithms.

Table of Contents

  1. Introduction
  2. Contributions
  3. Organization of the paper.
  4. Related Literature
  5. Model
  6. Problem formulation
  7. Dating Market.
  8. Displays, Likes and Dislikes.
  9. Dynamics.
  10. Matches.
  11. History.
  12. Platform's Goal and Design.
  13. Notation.
  14. One-lookahead Policies
  15. Integral Dating Heuristic
  16. ...and 33 more sections

Key Result

Proposition 3.4

The function $f^{t+1}(R,\mathbf{x}^t,\mathbf{w}^t)$ defined in eq: general next problem is not submodular in $R$.

Figures (5)

  • Figure 1: Overall Matches varying History Effect
  • Figure 2: Like Rates for Initial and Backlog Interactions
  • Figure 3: Sensitivity to Magnitude of History Effect: Linear Case
  • Figure 4: Matches per Benchmark by Platform Design
  • Figure 5: Distribution of Like Probabilities

Theorems & Definitions (62)

  • Remark 3.1
  • Remark 3.3
  • Proposition 3.4
  • Lemma 4.1
  • Theorem 4.3
  • Theorem 4.4
  • Definition 5.1
  • Lemma 5.2
  • Theorem 5.3
  • Theorem 5.4
  • ...and 52 more