Matching of Users and Creators in Two-Sided Markets with Departures

Daniel Huttenlocher,Hannah Li,Liang Lyu,Asuman Ozdaglar,James Siderius

2024-01-20

Abstract:Many online platforms of today, including social media sites, are two-sided markets bridging content creators and users. Most of the existing literature on platform recommendation algorithms largely focuses on user preferences and decisions, and does not simultaneously address creator incentives. We propose a model of content recommendation that explicitly focuses on the dynamics of user-content matching, with the novel property that both users and creators may leave the platform permanently if they do not experience sufficient engagement. In our model, each player decides to participate at each time step based on utilities derived from the current match: users based on alignment of the recommended content with their preferences, and creators based on their audience size. We show that a user-centric greedy algorithm that does not consider creator departures can result in arbitrarily poor total engagement, relative to an algorithm that maximizes total engagement while accounting for two-sided departures. Moreover, in stark contrast to the case where only users or only creators leave the platform, we prove that with two-sided departures, approximating maximum total engagement within any constant factor is NP-hard. We present two practical algorithms, one with performance guarantees under mild assumptions on user preferences, and another that tends to outperform algorithms that ignore two-sided departures in practice.

Computer Science and Game Theory,Machine Learning,Social and Information Networks,General Economics

What problem does this paper attempt to address?

The paper attempts to address the problem of how to design recommendation algorithms in bilateral markets (such as social media platforms) to optimize the matching between users and content creators, while considering the participation constraints of both parties. Specifically, most existing recommendation algorithms primarily focus on user preferences and decisions without simultaneously considering the incentive mechanisms for creators. This study proposes a recommendation model that explicitly focuses on the dynamics of matching between users and content creators and introduces the novelty that if users or creators fail to achieve sufficient engagement, they may permanently leave the platform.

The main contributions of the paper include:

1. **Proposing a new recommendation model**: This model considers the participation constraints of both users and creators, where users decide whether to continue using the platform based on the match between recommended content and their preferences, and creators decide whether to continue creating content based on their audience size.

2. **Analyzing the effects of different recommendation algorithms**: The paper demonstrates that a greedy algorithm based solely on user preferences can lead to a significant decline in total engagement if the departure of creators is not considered. In contrast, a forward-looking algorithm can maximize long-term total engagement by considering the participation constraints of both parties.

3. **Proving the complexity of the problem**: When considering the dynamic exit of both users and creators, approximating the problem of maximizing total engagement is proven to be NP-hard. This indicates that more complex algorithms are needed when addressing recommendation problems in bilateral markets.

4. **Proposing practical algorithms**: Based on the theoretical results mentioned above, the paper proposes two practical algorithms—Local Clustering and Creator Ranking with Potential Audience Size. These algorithms perform well in practice, especially in scenarios where user-centric algorithms are less effective.

In summary, this paper aims to improve overall engagement in bilateral markets by enhancing recommendation algorithms, particularly by optimizing strategies considering the potential exit of users and creators due to dissatisfaction.