@inproceedings{rodriguez-graff-2026-team,
    title = "Team Vivek Dhayaal at {S}em{E}val-2026 Task 13 Subtask {B}: Multi-Class Authorship Detection",
    author = "Rodriguez, David  and
      Graff, Mario",
    editor = "Kochmar, Ekaterina  and
      Ghosh, Debanjan  and
      North, Kai  and
      Komachi, Mamoru",
    booktitle = "Proceedings of the 20th {I}nternational {W}orkshop on {S}emantic {E}valuation (2026)",
    month = jul,
    year = "2026",
    address = "San Diego, California, USA",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl-workshops/2026.semeval-1.434/",
    pages = "3520--3523",
    ISBN = "979-8-89176-414-9",
    abstract = "This paper describes the system for SemEval-2026 Task 10 Subtask 2 on conspiracy detection. We explore a progressive modeling strategy comparing traditional lexical representations with contextual transformer models. Lexical baselines include Bag-of-Words and TF-IDF features combined with Logistic Regression and Ridge classifiers. We then fine-tune a DistilRoBERTa transformer model for binary classification.All experiments were conducted using only the official task data in a CPU-only environment without external datasets or data augmentation. Our objective was to achieve acceptable performance while minimizing computational resources and model complexity. Results show that the transformer model improves the best lexical baseline from 0.67 to 0.75. The work highlights that competitive performance in conspiracy detection can be obtained with lightweight and reproducible configurations."
}