Conflict detection for smart cities services: Poster abstract

The effort of trying to integrate individual services leads to the increasing potential conflicts across smart city services. This kind of conflicts might cause unsafe environment or degrade the overall performance. In this paper, we mainly focus on conflict detection by exploring the fundamental correlations among services and intend to monitor the city's safety and performance collectively. The key insight is that there is a multi-dimensional dependency in predicted conflicts. Therefore, we propose a novel multitask learning framework to detect various conflicts simultaneously. We evaluate this service conflict detection model with a large-scale real dataset across a few transportation and environment services, including the location and transaction information from the taxi, bus and subway systems. The results show the advantage of our system.