VSE knjižnice (vzajemna bibliografsko-kataložna baza podatkov COBIB.SI)
  • IJS@LT-EDI [Elektronski vir] : ensemble approaches to detect signs of depression from social media texts
    Caporusso, Jaya ; Hong Hanh, Tran Thi ; Pollak, Senja, 1980-
    This paper presents our ensembling solutions for detecting signs of depression in social me- dia text, as part of the Shared Task at LT- EDI@RANLP 2023. By leveraging social me- dia posts in English, ... the task involves the de- velopment of a system to accurately classify them as presenting signs of depression of one of three levels: “severe”, “moderate”, and “not depressed”. We verify the hypothesis that com- bining contextual information from a language model with local domain-specific features can improve the classifier’s performance. We do so by evaluating: (1) two global classifiers (sup- port vector machine and logistic regression); (2) contextual information from language models; and (3) the ensembling results. The best results were not achieved by any of the ensembling approaches, but by employing the RoBERTa language model.
    Vir: LTEDI 2023, Third Workshop on Language Technology for Equality, Diversity and Inclusion, September 7, 2023 [Elektronski vir] : proceedings of the third workshop on language technology for equality, diversity and inclusion (Str. 172-178)
    Vrsta gradiva - prispevek na konferenci ; neleposlovje za odrasle
    Leto - 2023
    Jezik - angleški
    COBISS.SI-ID - 166381571

    Povezava(-e):

    https://sites.google.com/view/lt-edi-2023/proceedings

Vnos na polico