logo Idiap Research Institute        
 [BibTeX] [Marc21]
Two Simple and Domain-independent Approaches for Early Detection of Anorexia
Type of publication: Book chapter
Citation: Burdisso_SPRINGER_2022
Booktitle: Early Detection of Mental Health Disorders by Social Media Monitoring: The First Five Years of the eRisk Project
Edition: 202
Year: 2022
Pages: 159-182
Publisher: Springer International Publishing
ISBN: 978-3-031-04431-1
URL: https://doi.org/10.1007/978-3-...
DOI: 10.1007/978-3-031-04431-1_7
Abstract: In this chapter, we describe the participation of our research team in the eRisk addressing the two editions of the early anorexia detection task. We used two domain-independent approaches to address this task. The first approach is based on a temporal-aware document representation, whereas the second one consists of a simple, interpretable, and novel text classification model specially designed for addressing early risk detection scenarios. Regarding the obtained results, in the first edition, we achieved the best ERDE5 value among all participant models using the first approach, whereas with the second one, the best precision (0.91). Besides, using the latter approach, in the second edition, we were able to achieve the best values for both ERDE5 and ERDE50, and also promising results in terms of the ranking-based metrics, obtaining the best values, consistently, across all four rankings.
Projects Idiap
Authors Burdisso, Sergio
Cagnina, Leticia
Errecalde, Marcelo
Montes-y-Gómez, Manuel
Added by: [UNK]
Total mark: 0
  • Burdisso_SPRINGER_2022.pdf