Social media platforms, such as Twitter, can be used to extract informationrelated to traffic events. Previous works focused mainly on classifying tweetsinto predefined categories (i.e., traffic or non-traffic) without many details oftraffic events. However, extracting traffic-related fine-grained information fromtweets is essential to build an intelligent transportation system. In this work,we address for the first time the problem of detecting traffic events using Twitter as two subtasks: (i) identifying whether a tweet is traffic-related or not asa text classification subtask, and (ii) extracting more fine-grained information(i.e., “what”, “when”, “where”, and the “consequence” of the traffic event) asa slot filling subtask. We also publish two Dutch Traffic Twitter datasets fromBelgium and the Brussels capital region. We propose using deep learning basedmethods that process the two subtasks separately or jointly. Experimental results indicate that the proposed architectures achieve high performance scores(i.e., more than 95% F1 score) on the constructed datasets for both subtasks,even in a transfer learning scenario. In addition, incorporating tweet-level information in each of the tokens comprising the tweet (for the BERT-based model) can lead to a performance improvement for the joint setting. Our datasets and code are available on GitHub.
Yang, X, Bekoulis, I & Deligiannis, N 2023, 'Traffic Event Detection as a Slot Filling Problem', Engineering Applications of Artificial Intelligence, vol. 123, no. PA, 106202, pp. 1-13. https://doi.org/10.1016/j.engappai.2023.106202
Yang, X., Bekoulis, I., & Deligiannis, N. (2023). Traffic Event Detection as a Slot Filling Problem. Engineering Applications of Artificial Intelligence, 123(PA), 1-13. Article 106202. https://doi.org/10.1016/j.engappai.2023.106202
@article{5832fbd8816549baaa0a453b3541231d,
title = "Traffic Event Detection as a Slot Filling Problem",
abstract = "Social media platforms, such as Twitter, can be used to extract informationrelated to traffic events. Previous works focused mainly on classifying tweetsinto predefined categories (i.e., traffic or non-traffic) without many details oftraffic events. However, extracting traffic-related fine-grained information fromtweets is essential to build an intelligent transportation system. In this work,we address for the first time the problem of detecting traffic events using Twitter as two subtasks: (i) identifying whether a tweet is traffic-related or not asa text classification subtask, and (ii) extracting more fine-grained information(i.e., “what”, “when”, “where”, and the “consequence” of the traffic event) asa slot filling subtask. We also publish two Dutch Traffic Twitter datasets fromBelgium and the Brussels capital region. We propose using deep learning basedmethods that process the two subtasks separately or jointly. Experimental results indicate that the proposed architectures achieve high performance scores(i.e., more than 95% F1 score) on the constructed datasets for both subtasks,even in a transfer learning scenario. In addition, incorporating tweet-level information in each of the tokens comprising the tweet (for the BERT-based model) can lead to a performance improvement for the joint setting. Our datasets and code are available on GitHub.",
author = "Xiangyu Yang and Ioannis Bekoulis and Nikos Deligiannis",
note = "Funding Information: This work has been supported in part by the Innoviris Project MobiPulse (2018-EXPLORE-22a) and in part by the Flemish Government, under the “Onderzoeksprogramma Artifici{\"e}le Intelligentie (AI) Vlaanderen” programme. Publisher Copyright: {\textcopyright} 2023 Elsevier Ltd Copyright: Copyright 2023 Elsevier B.V., All rights reserved.",
year = "2023",
month = aug,
doi = "10.1016/j.engappai.2023.106202",
language = "English",
volume = "123",
pages = "1--13",
journal = "Engineering Applications of Artificial Intelligence",
issn = "0952-1976",
publisher = "Elsevier",
number = "PA",
}