Spatial and Temporal Pricing Approach for Tasks in Spatial Crowdsourcing

摘要

Pricing is an important issue in spatial crowdsourcing (SC). Current pricing mechanisms are usually built on online learning algorithms, so they fail to capture the dynamics of users' price preference timely. In this paper, we focus on the pricing for task requesters with the goal of maximizing the total revenue gained by the SC platform. By considering the relationship between the price and the task, space, and time, a spatial and temporal pricing framework based task-transaction history is proposed. We model the price of a task as a three-dimensional tensor (task-space-time) and complete the missing entries with the assistant of historical data and other three context matrices. We conduct extensive experiments on a real taxi-hailing dataset. The experimental results show the effectiveness of the proposed pricing framework.