CRII: RI: Fairness and Profitability in Online Matching Markets

点击翻译按钮获取中文摘要

Matching markets involve heterogeneous agents (typically from two parties) who are paired for mutual benefits. During the last decade, matching markets have emerged and proliferated through the medium of the Internet. They have evolved into a new style, called Online Matching Markets (OMMs), with examples including ridesharing, crowdsourcing, and Internet advertising. OMMs feature dynamic arrivals of agents and the real-time decision-making requirement. For ridesharing, a wide range of gender- and race-based discrimination has been reported from drivers to riders. In particular, it has been shown that riders from minority groups have received unfair cancellations from drivers much more frequently than others. In spatial crowdsourcing, reports mention that workers tried to avoid tasks from socioeconomically disadvantaged places. This project aims to leverage powerful algorithmic tools to mitigate discrimination and promote revenue for a large variety of real-world OMMs, including ridesharing and crowdsourcing. The research will cover the following three directions. First, the project will try to carefully re-craft the matching policy in ridesharing such that it is favorable for disadvantaged groups who are vulnerable to discrimination. Second, this project will seek to quantitatively study the inherent relationship between fairness and revenue achieved by various practical OMMs. The two objectives--high fairness and high revenue--can be somewhat conflicting with each other. The project will develop multi-objective optimization models to investigate the tradeoff between profitability and equity. Third, this project aims to leverage the arrival patterns of online workers to increase revenue in crowdsourcing. Two prominent challenging issues in crowdsourcing are workers' dynamic arrivals and their unknown skills for tasks. The project will utilize machine-learning techniques to learn the arrival patterns of online workers and use online-matching based models to harvest those patterns. The proposed research could have significant broader impacts on society as it can potentially contribute to reducing discrimination in ridesharing and crowdsourcing.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.