Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Welfare-Optimized Recommender Systems

Benjamin Heymann, Flavian Vasile, David Rohde

TL;DR

A recommender system based on the Random Utility Model, which unify the retailer and shoppers' contradictory objectives into a single welfare metric, opens the door to Welfare-Optimized Recommender Systems, couponing, and price optimization.

Abstract

We present a recommender system based on the Random Utility Model. Online shoppers are modeled as rational decision makers with limited information, and the recommendation task is formulated as the problem of optimally enriching the shopper's awareness set. Notably, the price information and the shopper's Willingness-To-Pay play crucial roles. Furthermore, to better account for the commercial nature of the recommendation, we unify the retailer and shoppers' contradictory objectives into a single welfare metric, which we propose as a new recommendation goal. We test our framework on synthetic data and show its performance in a wide range of scenarios. This new framework, that was absent from the Recommender System literature, opens the door to Welfare-Optimized Recommender Systems, couponing, and price optimization.

Welfare-Optimized Recommender Systems

TL;DR

A recommender system based on the Random Utility Model, which unify the retailer and shoppers' contradictory objectives into a single welfare metric, opens the door to Welfare-Optimized Recommender Systems, couponing, and price optimization.

Abstract

We present a recommender system based on the Random Utility Model. Online shoppers are modeled as rational decision makers with limited information, and the recommendation task is formulated as the problem of optimally enriching the shopper's awareness set. Notably, the price information and the shopper's Willingness-To-Pay play crucial roles. Furthermore, to better account for the commercial nature of the recommendation, we unify the retailer and shoppers' contradictory objectives into a single welfare metric, which we propose as a new recommendation goal. We test our framework on synthetic data and show its performance in a wide range of scenarios. This new framework, that was absent from the Recommender System literature, opens the door to Welfare-Optimized Recommender Systems, couponing, and price optimization.
Paper Structure (35 sections, 7 equations, 2 figures, 4 tables)

This paper contains 35 sections, 7 equations, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Recommender Systems
  3. Shortcomings of current approaches
  4. Our approach
  5. Paper contents
  6. Related Work
  7. Performance-optimized Recommendation
  8. Price in Recommendation
  9. Recommendation and Econometric Models
  10. Random Utility Models
  11. Welfare and Recommendation
  12. A shopper's behavioural model
  13. User Decision Model
  14. Random Utility Maximization/ Matrix Factorization (RUM-MF)
  15. Gumbel noise
  16. ...and 20 more sections

Figures (2)

  • Figure 1: The model of the user as a rational decision-maker with a limited awareness set. At time t1 and t2, the user observes different models of phones, with their prices. At time t3, they decide whether to buy one of the two phones, or to leave the website empty-handed. This decision is the output of a rational process of utility maximization that takes into account both the features and the prices of the two phones. The novelty of our framework is that we make explicit the difference between what the user knows (the awareness set) and the entire catalog.
  • Figure 2: The role of recommendation in a limited information setup. Based on its knowledge of the user, the RS points the user to a new item that is likely to be useful. The value of RS is equal to the difference in utility between the user's decisions with or without the recommendation.