@inproceedings{bayram-benhiba-2021-determining,
    title = "Determining a Person`s Suicide Risk by Voting on the Short-Term History of Tweets for the {CLP}sych 2021 Shared Task",
    author = "Bayram, Ulya  and
      Benhiba, Lamia",
    editor = "Goharian, Nazli  and
      Resnik, Philip  and
      Yates, Andrew  and
      Ireland, Molly  and
      Niederhoffer, Kate  and
      Resnik, Rebecca",
    booktitle = "Proceedings of the Seventh Workshop on Computational Linguistics and Clinical Psychology: Improving Access",
    month = jun,
    year = "2021",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/jlcl-multiple-ingestion/2021.clpsych-1.8/",
    doi = "10.18653/v1/2021.clpsych-1.8",
    pages = "81--86",
    abstract = "In this shared task, we accept the challenge of constructing models to identify Twitter users who attempted suicide based on their tweets 30 and 182 days before the adverse event`s occurrence. We explore multiple machine learning and deep learning methods to identify a person`s suicide risk based on the short-term history of their tweets. Taking the real-life applicability of the model into account, we make the design choice of classifying on the tweet level. By voting the tweet-level suicide risk scores through an ensemble of classifiers, we predict the suicidal users 30-days before the event with an 81.8{\%} true-positives rate. Meanwhile, the tweet-level voting falls short on the six-month-long data as the number of tweets with weak suicidal ideation levels weakens the overall suicidal signals in the long term."
}