KUET at StanceNakba Shared Task: StanceMoE: Mixture-of-Experts Architecture for Stance Detection
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911559934017536 |
|---|---|
| author | Shafi, Abdullah Al Islam, Md. Milon Hossain, Sk. Imran Hasan, K. M. Azharul |
| author_facet | Shafi, Abdullah Al Islam, Md. Milon Hossain, Sk. Imran Hasan, K. M. Azharul |
| contents | Actor-level stance detection aims to determine an author expressed position toward specific geopolitical actors mentioned or implicated in a text. Although transformer-based models have achieved relatively good performance in stance classification, they typically rely on unified representations that may not sufficiently capture heterogeneous linguistic signals, such as contrastive discourse structures, framing cues, and salient lexical indicators. This motivates the need for adaptive architectures that explicitly model diverse stance-expressive patterns. In this paper, we propose StanceMoE, a context-enhanced Mixture-of-Experts (MoE) architecture built upon a fine-tuned BERT encoder for actor-level stance detection. Our model integrates six expert modules designed to capture complementary linguistic signals, including global semantic orientation, salient lexical cues, clause-level focus, phrase-level patterns, framing indicators, and contrast-driven discourse shifts. A context-aware gating mechanism dynamically weights expert contributions, enabling adaptive routing based on input characteristics. Experiments are conducted on the StanceNakba 2026 Subtask A dataset, comprising 1,401 annotated English texts where the target actor is implicit in the text. StanceMoE achieves a macro-F1 score of 94.26%, outperforming traditional baselines, and alternative BERT-based variants. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_00878 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | KUET at StanceNakba Shared Task: StanceMoE: Mixture-of-Experts Architecture for Stance Detection Shafi, Abdullah Al Islam, Md. Milon Hossain, Sk. Imran Hasan, K. M. Azharul Computation and Language Artificial Intelligence Machine Learning Actor-level stance detection aims to determine an author expressed position toward specific geopolitical actors mentioned or implicated in a text. Although transformer-based models have achieved relatively good performance in stance classification, they typically rely on unified representations that may not sufficiently capture heterogeneous linguistic signals, such as contrastive discourse structures, framing cues, and salient lexical indicators. This motivates the need for adaptive architectures that explicitly model diverse stance-expressive patterns. In this paper, we propose StanceMoE, a context-enhanced Mixture-of-Experts (MoE) architecture built upon a fine-tuned BERT encoder for actor-level stance detection. Our model integrates six expert modules designed to capture complementary linguistic signals, including global semantic orientation, salient lexical cues, clause-level focus, phrase-level patterns, framing indicators, and contrast-driven discourse shifts. A context-aware gating mechanism dynamically weights expert contributions, enabling adaptive routing based on input characteristics. Experiments are conducted on the StanceNakba 2026 Subtask A dataset, comprising 1,401 annotated English texts where the target actor is implicit in the text. StanceMoE achieves a macro-F1 score of 94.26%, outperforming traditional baselines, and alternative BERT-based variants. |
| title | KUET at StanceNakba Shared Task: StanceMoE: Mixture-of-Experts Architecture for Stance Detection |
| topic | Computation and Language Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2604.00878 |