Paper Number

2529

Paper Type

Complete

Description

Online entertainment platforms such as Youtube host a vast amount of user-generated content (UGC). The unique feature of two-sided UGC entertainment platforms is that creators’ content generation and users’ content usage can influence each other. However, traditional recommender systems often emphasize content usage but ignore content generation, leading to a misalignment between these two goals. To address the challenge, this paper proposes a prescriptive uplift framework to balance content generation and usage through personalized recommendations. Specifically, we first predict the heterogeneous treatment effects (HTEs) of recommended contents on creators’ content generation and users’ content usage, then consider these two predicted HTEs simultaneously in an optimization model to determine the recommended contents for each user. Using a large-scale real-world dataset, we demonstrate that the proposed recommendation method better balances content generation and usage and brings a 42% increase in participants’ activity compared to existing benchmark methods.

Comments

20-OnlineReviews

Share

COinS
Best Paper Nominee badge
 
Dec 12th, 12:00 AM

Personalized Recommendation for Balancing Content Generation and Usage on Two-Sided Entertainment Platforms

Online entertainment platforms such as Youtube host a vast amount of user-generated content (UGC). The unique feature of two-sided UGC entertainment platforms is that creators’ content generation and users’ content usage can influence each other. However, traditional recommender systems often emphasize content usage but ignore content generation, leading to a misalignment between these two goals. To address the challenge, this paper proposes a prescriptive uplift framework to balance content generation and usage through personalized recommendations. Specifically, we first predict the heterogeneous treatment effects (HTEs) of recommended contents on creators’ content generation and users’ content usage, then consider these two predicted HTEs simultaneously in an optimization model to determine the recommended contents for each user. Using a large-scale real-world dataset, we demonstrate that the proposed recommendation method better balances content generation and usage and brings a 42% increase in participants’ activity compared to existing benchmark methods.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.