Multi-task Learning for Transit Service Disruption Detection

Published in IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), 2018

Abstract-With the rapid growth in urban transit networks inrecent years, detecting service disruptions in a timely manneris a problem of increased interest to service providers. Transit agencies are seeking to move beyond traditional customerquestionnaires and manual service inspections to leveraging opensource indicators like social media for deteting emerging transitevents. In this paper, we leverage Twitter data for early detectionof metro service disruptions. Inspired by the multi-task learningframework, we propose the Metro Disruption Detection Model, which captures the semantic similarity between transit lines in Twitter space. We propose novel constraints on feature semanticsimilarity exploiting prior knowledge about the spatial connectivity and shared tracks of the metro network. An algorithmbased on the alternating direction method of multipliers (ADMM) framework is developed to solve the proposed model. We runextensive experiments and comparisons to other models withreal world Twitter data and transit disruption records from the Washington Metropolitan Area Transit Authority (WMATA) to justify the efficacy of our model.

Download paper here

Recommended citation:

@inproceedings{ji2018multi,
    title={Multi-task Learning for Transit Service Disruption Detection},
    author={Ji, Taoran and Fu, Kaiqun and Self, Nathan and Lu, Chang-Tien and Ramakrishnan, Naren},
    booktitle={2018 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM)},
    pages={634--641},
    year={2018},
    organization={IEEE}
}