UCSC NLP at SemEval-2026 Task 10: Boundary-Aware Span Extraction and RoBERTa Classification for Conspiracy Detection

Dom Marhoefer; Milos Suvakovic; Glenn Grant-Richards; Aidan Pinero; Ryan King

UCSC NLP at SemEval-2026 Task 10: Boundary-Aware Span Extraction and RoBERTa Classification for Conspiracy Detection

Computation and Language 2026-07-06 v1 Social and Information Networks

Authors: Dom Marhoefer , Milos Suvakovic , Glenn Grant-Richards , Aidan Pinero , Ryan King

Abstract

We present our systems for SemEval-2026 Task 10 (PsyCoMark), addressing conspiracy marker extraction (Subtask 1) and document-level conspiracy detection (Subtask 2). For marker extraction, we formulate the task as multi-label span classification over enumerated candidate spans, using IoU >= 0.95 positive labeling, hard-negative sampling, and containment-based non-maximum suppression (NMS) with boundary-aware span representations. Document classification is modeled independently using a sequence classifier with label smoothing and a stratified train-validation split. Analysis shows that entity-like roles (Actor, Victim) are detected robustly, while abstract roles (Action, Effect, Evidence) remain sensitive to boundary criteria. On the official test set, our systems rank 7th in Subtask 1 (0.2251 macro F1) and 11th in Subtask 2 (0.7694 weighted F1).

Keywords

conspiracy detection span extraction boundary-aware representation multi-label classification RoBERTa SemEval

Cite

@article{arxiv.2607.05689,
  title  = {UCSC NLP at SemEval-2026 Task 10: Boundary-Aware Span Extraction and RoBERTa Classification for Conspiracy Detection},
  author = {Dom Marhoefer and Milos Suvakovic and Glenn Grant-Richards and Aidan Pinero and Ryan King},
  journal= {arXiv preprint arXiv:2607.05689},
  year   = {2026}
}

Comments

6 pages, 2 tables. System description paper for SemEval-2026 Task 10 (PsyCoMark: Psycholinguistic Conspiracy Marker Extraction and Detection)