Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Monotonic Modeling with Spatio-Temporal Adaptive Awareness in Diverse Marketing

Bin Li, Jiayan Pei, Feiyang Xiao, Yifan Zhao, Zhixing Zhang, Diwei Liu, HengXu He, Jia Jia

TL;DR

This paper tackles budget-efficient incentive allocation for Online Food Ordering Services by modeling monotonic user responses that vary across time and location. It introduces CoMAN, a Constrained Monotonic Adaptive Network that fuses two spatio-temporal perception modules with a CMNN-based monotonic layer and a flexible Four-Parameter Model (FPM) built from Convex CLU activations to learn a strictly monotonic incentive-sensitivity function conditioned on $time$ and $place$. The method combines Adaptive Activation and Spatio-temporal Attention to dynamically adjust convexity/concavity and function expression, achieving accurate incentive-response modeling under budget constraint $\mathcal{B}$. Extensive offline and online experiments on Ele.me campaigns show CoMAN outperforming state-of-the-art baselines in predictive accuracy and distributional similarity, while enabling more efficient subsidy use and improved business metrics such as CVR, GMV, and orders. These results suggest practical impact for scalable, region-aware pricing in mobile marketing.

Abstract

In the mobile internet era, the Online Food Ordering Service (OFOS) emerges as an integral component of inclusive finance owing to the convenience it brings to people. OFOS platforms offer dynamic allocation incentives to users and merchants through diverse marketing campaigns to encourage payments while maintaining the platforms' budget efficiency. Despite significant progress, the marketing domain continues to face two primary challenges: (i) how to allocate a limited budget with greater efficiency, demanding precision in predicting users' monotonic response (i.e. sensitivity) to incentives, and (ii) ensuring spatio-temporal adaptability and robustness in diverse marketing campaigns across different times and locations. To address these issues, we propose a Constrained Monotonic Adaptive Network (CoMAN) method for spatio-temporal perception within marketing pricing. Specifically, we capture spatio-temporal preferences within attribute features through two foundational spatio-temporal perception modules. To further enhance catching the user sensitivity differentials to incentives across varied times and locations, we design modules for learning spatio-temporal convexity and concavity as well as for expressing sensitivity functions. CoMAN can achieve a more efficient allocation of incentive investments during pricing, thus increasing the conversion rate and orders while maintaining budget efficiency. Extensive offline and online experimental results within our diverse marketing campaigns demonstrate the effectiveness of the proposed approach while outperforming the monotonic state-of-the-art method.

Enhancing Monotonic Modeling with Spatio-Temporal Adaptive Awareness in Diverse Marketing

TL;DR

This paper tackles budget-efficient incentive allocation for Online Food Ordering Services by modeling monotonic user responses that vary across time and location. It introduces CoMAN, a Constrained Monotonic Adaptive Network that fuses two spatio-temporal perception modules with a CMNN-based monotonic layer and a flexible Four-Parameter Model (FPM) built from Convex CLU activations to learn a strictly monotonic incentive-sensitivity function conditioned on and . The method combines Adaptive Activation and Spatio-temporal Attention to dynamically adjust convexity/concavity and function expression, achieving accurate incentive-response modeling under budget constraint . Extensive offline and online experiments on Ele.me campaigns show CoMAN outperforming state-of-the-art baselines in predictive accuracy and distributional similarity, while enabling more efficient subsidy use and improved business metrics such as CVR, GMV, and orders. These results suggest practical impact for scalable, region-aware pricing in mobile marketing.

Abstract

In the mobile internet era, the Online Food Ordering Service (OFOS) emerges as an integral component of inclusive finance owing to the convenience it brings to people. OFOS platforms offer dynamic allocation incentives to users and merchants through diverse marketing campaigns to encourage payments while maintaining the platforms' budget efficiency. Despite significant progress, the marketing domain continues to face two primary challenges: (i) how to allocate a limited budget with greater efficiency, demanding precision in predicting users' monotonic response (i.e. sensitivity) to incentives, and (ii) ensuring spatio-temporal adaptability and robustness in diverse marketing campaigns across different times and locations. To address these issues, we propose a Constrained Monotonic Adaptive Network (CoMAN) method for spatio-temporal perception within marketing pricing. Specifically, we capture spatio-temporal preferences within attribute features through two foundational spatio-temporal perception modules. To further enhance catching the user sensitivity differentials to incentives across varied times and locations, we design modules for learning spatio-temporal convexity and concavity as well as for expressing sensitivity functions. CoMAN can achieve a more efficient allocation of incentive investments during pricing, thus increasing the conversion rate and orders while maintaining budget efficiency. Extensive offline and online experimental results within our diverse marketing campaigns demonstrate the effectiveness of the proposed approach while outperforming the monotonic state-of-the-art method.
Paper Structure (14 sections, 8 theorems, 17 equations, 8 figures, 5 tables)

This paper contains 14 sections, 8 theorems, 17 equations, 8 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Preliminaries
  4. Methods
  5. Experiments
  6. Experimental Setup
  7. Experimental Results
  8. Conclusion
  9. Marketing Monotonic
  10. Detailed Proofs
  11. Monotone Universal Approximation
  12. Convex CLU
  13. Methodological Property
  14. Implemental Details

Key Result

Theorem 1

Let $\breve{\rho} \in \mathcal{\breve{A}}$. Then any multivariate continuous monotone function f on a compact subset of $\mathbb{R}^k$ can be approximated with a monotone constrained neural network of at most $k$ layers using $\rho$ as the activation function.

Figures (8)

  • Figure 1: Overview of our proposed Constrained Monotonic Adaptive Network (CoMAN) architecture.
  • Figure 2: Original Four-Parameter Model (FPM) and combined with Spatio-temporal Attention (S-t Attention) fitting.
  • Figure 3: CLU based activation functions construction
  • Figure 4: Visualization of the average redemption amount of coupons in the top five cities with the most users across different periods.
  • Figure 5: Visualization of the prediction scores for incentive responses from various online models and the corresponding ground truth. Results are observed overall, with concurrent evaluations segmented by periods and cities.
  • ...and 3 more figures

Theorems & Definitions (8)

  • Theorem 1
  • Theorem 2
  • Lemma 1
  • Lemma 2
  • Theorem 3
  • Lemma 3
  • Lemma 4
  • Corollary 1