Beyond Self-Interest: How Group Strategies Reshape Content Creation in Recommendation Platforms?

Authors: Yaolong Yu, Fan Yao, Sinno Jialin Pan

ICML 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Empirical results from simulations further support the effectiveness of the user engagement rewarding mechanism. We construct simulations to further validate the effectiveness of the user engagement rewarding mechanism.
Researcher Affiliation Academia 1Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong, China 2Department of Computer Science, University of Virginia, Charlottesville, Virginia, USA. Correspondence to: Sinno Jialin Pan <EMAIL>.
Pseudocode Yes Algorithm 1 Self-Greedy Algorithm for Constructing a PNE ... Algorithm 2 (LBR) Local Better Response update at time step t
Open Source Code No The paper does not explicitly state that source code for the described methodology is being released, nor does it provide a link to a code repository. It refers to algorithms from other works (Yao et al., 2024c;a) but not its own implementation code.
Open Datasets No Synthetic Environment For the synthetic environment, we first construct the user population as follows: we fix an embedding dimension d = 5 and independently sample m = 10 users from the unit sphere Sd 1. The distribution of the 10 users is p = 1 200 [100, 50, 20, 10, 10, 5, 2, 1, 1, 1] .
Dataset Splits No The paper describes constructing a 'Synthetic Environment' for simulations, where user populations and creator behaviors are modeled. It does not involve splitting a pre-existing dataset into training, validation, or test sets in the traditional sense, as it is a simulated setup rather than an experiment on an empirical dataset.
Hardware Specification No The paper does not provide specific details regarding the hardware (e.g., CPU, GPU models, memory, or computational clusters) used to perform the simulations or experiments.
Software Dependencies No The paper does not list specific software dependencies with version numbers (e.g., programming languages, libraries, or frameworks) used for implementing the simulations.
Experiment Setup Yes We set (β, K) = (0.1, 5) by default. This synthetic dataset characterizes a class of clustered user preference distributions, such as majority versus minority user groups. On the creators side, there are n = 30 creators in total, including one group of creators. We vary the group size nc {10, 15, 20, 25, 30}. The group of creators strategies is initialized near user x1, while the remaining creators strategies are initialized near the other users. The results are presented in Figure 1. ... The time horizon is set to T = 100. ... We evaluate K over the set {1, 2, 3, 6, 10, 15}. ... We test two values of β: β = 0.1 and β = 10.