MMDS-Bench is a new diagnostic benchmark for multimodal dynamic stance classification in social media parent‑reply interactions. It contains 3,482 multimodal instances annotated with a seven‑label stance taxonomy, plus an 800‑instance subset that demands structured reasoning over parent and reply understanding and stance‑relation inference. The benchmark also tags each instance with five challenge factors—multimodal fusion, parent framing, non‑literal expression, interaction reasoning, and label‑boundary ambiguity—and evaluates 12 multimodal large language models using a reference‑grounded LLM‑judge protocol, revealing that current models still struggle with relational inference beyond separate parent and reply comprehension.
arXiv:2607. 24191v1 Announce Type: cross Abstract: Conversational stance detection has shifted from static text analysis to dynamic multimodal modeling.
The paper investigates how dense embedding models can be used for stance-aware argument retrieval, a task that requires both topic relevance and correct stance (support or attack) toward a claim. Experiments reveal that current models favor topical overlap and ignore stance, and that contrastive training to fix this bias leads to over-correction, where models focus too much on polarity keywords at the expense of topic relevance. To address this, the authors propose diagnostic word-ablation metrics and a data‑centric solution involving a balanced argument curriculum and LLM‑augmented stance‑inverted arguments, which helps powerful models learn deeper directional logic and improves stance‑aware retrieval performance.
arXiv:2604. 00878v2 Announce Type: replace-cross Abstract: Actor-level stance detection aims to determine an author expressed position toward specific geopolitical actors mentioned or implicated in a text.
MoPrO‑SD is a stance‑specific prompt optimization framework that breaks prompts into semantic modules and refines them sequentially with LLM feedback. It evaluates three refinement strategies—instruction‑only, instruction with reasoning, and instruction with per‑example reasoning—using class‑conditioned errors to improve stance detection. The authors also release VaccineStance, a new human‑annotated benchmark for vaccine stance detection, and demonstrate that MoPrO‑SD outperforms human‑written base prompts across multiple datasets and open‑weight LLMs.
arXiv:2608.29066v1 Announce Type: cross Abstract: Stance detection is crucial for understanding the underlying attitude of an expression towards a target. Conversational stance detection is a more ch...
arXiv:2606. 15694v1 Announce Type: cross Abstract: Multimodal large language models (MLLMs) have demonstrated remarkable capabilities in understanding complex multimodal content.
arXiv:2606. 10194v1 Announce Type: cross Abstract: Climate change research increasingly requires AI systems that reason across text, dynamic visual content, and scientific figures, yet existing climate QA benchmarks are small, mostly textual, and cover a narrow range of models.
arXiv:2606.06443v3 Announce Type: replace Abstract: Large language models are increasingly used to simulate social media users and infer how individuals may respond to online discussions. However, it...
arXiv:2608. 10810v1 Announce Type: cross Abstract: Emotion understanding in discourse requires reasoning beyond surface sentiment because speakers often convey affect through indirect, implicit, polite, ironic, or deliberately mismatched expressions.
The paper introduces VFStance, a method that uses image generation to make implicit stance cues in news articles more explicit through visual framing. It targets article-level news stance detection, a task complicated by subtle, structurally complex texts. Experiments show VFStance outperforms existing methods, and a user study with 200 participants demonstrates that the visual framing makes stance signals more noticeable in a snippet-based news consumption setting.
The paper introduces a stance‑granularity selection policy for dialog systems that engage users with alternative viewpoints. By using Perspectivized Stance Vectors to break positions into concept‑level stances, the system selects the most novel viewpoint within a bandit‑controlled distance bound. Experiments with 63 participants show that concept‑level stance yields more varied exposure sequences and broader coverage of topic positions, though it does not affect depth of justification or reconstruction accuracy.
M$^3$R-Bench is a new benchmark comprising 1,000 human‑annotated image‑text pairs that jointly label metaphor occurrence, Target‑Source mapping, and sentiment, and provide step‑by‑step explanations for evidence identification, mapping, and sentiment inference. Existing models often rely on textual shortcuts and produce mappings that conflict with visual evidence, as shown by evaluations on this dataset. To address this, the authors propose M$^3$R‑Reasoner, a model that uses curriculum‑based reasoning supervision and task‑aware reinforcement learning, achieving a 37.90% Joint Label Accuracy and outperforming both proprietary baselines and GPT‑5.5 on evidence and sentiment justification.