Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Pareto-based Multi-Objective Recommender System with Forgetting Curve

Jipeng Jin, Zhaoxiang Zhang, Zhiheng Li, Xiaofeng Gao, Xiongwen Yang, Lei Xiao, Jie Jiang

TL;DR

PMORS addresses negative feedback in cascaded recommender systems by modeling recency with the forgetting curve and balancing it with utility via a Pareto optimization framework. It introduces two objectives, $L_{LTR}$ for ranking consistency and $L_{FG}$ for recency-driven penalties, and jointly optimizes them through a Pareto solver that computes adaptive weights $\alpha$ to reach Pareto efficiency. The framework combines a pre-ranking model, a ranking model, and a forgetting mechanism, and demonstrates effectiveness on public video data and in a production deployment on WeChat Channels, achieving up to a GMV uplift of $+1.45\%$. The work contributes the first integration of the Ebbinghaus forgetting curve into a multi-objective recommender and provides a scalable solver for dynamic trade-offs between recency and utility in production settings.

Abstract

Recommender systems with cascading architecture play an increasingly significant role in online recommendation platforms, where the approach to dealing with negative feedback is a vital issue. For instance, in short video platforms, users tend to quickly slip away from candidates that they feel aversive, and recommender systems are expected to receive these explicit negative feedbacks and make adjustments to avoid these recommendations. Considering recency effect in memories, we propose a forgetting model based on Ebbinghaus Forgetting Curve to cope with negative feedback. In addition, we introduce a Pareto optimization solver to guarantee a better trade-off between recency and model performance. In conclusion, we propose Pareto-based Multi-Objective Recommender System with forgetting curve (PMORS), which can be applied to any multi-objective recommendation and show sufficiently superiority when facing explicit negative feedback. We have conducted evaluations of PMORS and achieved favorable outcomes in short-video scenarios on both public dataset and industrial dataset. After being deployed on an online short video platform named WeChat Channels in May, 2023, PMORS has not only demonstrated promising results for both consistency and recency but also achieved an improvement of up to +1.45% GMV.

Pareto-based Multi-Objective Recommender System with Forgetting Curve

TL;DR

PMORS addresses negative feedback in cascaded recommender systems by modeling recency with the forgetting curve and balancing it with utility via a Pareto optimization framework. It introduces two objectives, for ranking consistency and for recency-driven penalties, and jointly optimizes them through a Pareto solver that computes adaptive weights to reach Pareto efficiency. The framework combines a pre-ranking model, a ranking model, and a forgetting mechanism, and demonstrates effectiveness on public video data and in a production deployment on WeChat Channels, achieving up to a GMV uplift of . The work contributes the first integration of the Ebbinghaus forgetting curve into a multi-objective recommender and provides a scalable solver for dynamic trade-offs between recency and utility in production settings.

Abstract

Recommender systems with cascading architecture play an increasingly significant role in online recommendation platforms, where the approach to dealing with negative feedback is a vital issue. For instance, in short video platforms, users tend to quickly slip away from candidates that they feel aversive, and recommender systems are expected to receive these explicit negative feedbacks and make adjustments to avoid these recommendations. Considering recency effect in memories, we propose a forgetting model based on Ebbinghaus Forgetting Curve to cope with negative feedback. In addition, we introduce a Pareto optimization solver to guarantee a better trade-off between recency and model performance. In conclusion, we propose Pareto-based Multi-Objective Recommender System with forgetting curve (PMORS), which can be applied to any multi-objective recommendation and show sufficiently superiority when facing explicit negative feedback. We have conducted evaluations of PMORS and achieved favorable outcomes in short-video scenarios on both public dataset and industrial dataset. After being deployed on an online short video platform named WeChat Channels in May, 2023, PMORS has not only demonstrated promising results for both consistency and recency but also achieved an improvement of up to +1.45% GMV.
Paper Structure (25 sections, 11 equations, 7 figures, 4 tables, 2 algorithms)

This paper contains 25 sections, 11 equations, 7 figures, 4 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Works
  3. Methodology
  4. Preliminary
  5. Pareto Theory
  6. Ebbinghaus Forgetting Curve
  7. Proposed Framework
  8. Ranking and Pre-ranking Model
  9. Forgetting Model
  10. Pareto Solver
  11. Training Strategy
  12. Experiments
  13. Experiment Setup
  14. Dataset
  15. Evaluation Metrics
  16. ...and 10 more sections

Figures (7)

  • Figure 1: The Process of A Typical Recommender System
  • Figure 2: (a) shows the structure of traditional LTR, consisting of a pre-ranking model (orange) and a fixed pretrained ranking model (green). (b) illustrates the forgetting model (blue) which is used to estimate a new objective called recency and a Pareto optimization solver which aggregates the two losses and purposes preference weights that lead the model to Pareto efficiency.
  • Figure 3: Overview Structure of Forgetting Model
  • Figure 4: Scalability Results
  • Figure 5: Consistency and Recency in Ablation Study
  • ...and 2 more figures

Theorems & Definitions (3)

  • definition 1: Pareto Dominance
  • definition 2: Pareto Optimality/Pareto Efficiency
  • definition 3: Pareto Stationarity