Hailiang Zhu | Searcharxiv

arXiv2026

CommSense: Facilitating Bias-Aware and Reflective Navigation of Online Comments for Rational Judgment

Online comments significantly influence users' judgments, yet their presentation, often determined by platform algorithms, can introduce biases, such as anchoring effects, which distort reasoning. While existing research emphasizes mitigating individual cognitive biases, the evolution of user judgments during comment engagement remains overlooked. This study investigates how presentation cues impact reasoning and explores interface design strategies to mitigate bias. Through a preliminary experiment (N=18) and a co-design workshop, we identified key challenges users face across a four-stage process and distilled four design requirements: pre-engagement framing, interactive organization, reflective prompts, and synthesis support. Based on these insights, we developed CommSense, an on-the-fly plugin that enhances user engagement with online comments by providing visual overviews and lightweight prompts to guide reasoning. A between-subject evaluation (N=24) demonstrates that CommSense improves bias awareness and reflective thinking, helping users produce more comprehensive, evidence-based rationales while maintaining high usability.

cs.HC↗

arXiv2025

ROADWork: A Dataset and Benchmark for Learning to Recognize, Observe, Analyze and Drive Through Work Zones

Perceiving and autonomously navigating through work zones is a challenging and underexplored problem. Open datasets for this long-tailed scenario are scarce. We propose the ROADWork dataset to learn to recognize, observe, analyze, and drive through work zones. State-of-the-art foundation models fail when applied to work zones. Fine-tuning models on our dataset significantly improves perception and navigation in work zones. With ROADWork dataset, we discover new work zone images with higher precision (+32.5%) at a much higher rate (12.8$\times$) around the world. Open-vocabulary methods fail too, whereas fine-tuned detectors improve performance (+32.2 AP). Vision-Language Models (VLMs) struggle to describe work zones, but fine-tuning substantially improves performance (+36.7 SPICE). Beyond fine-tuning, we show the value of simple techniques. Video label propagation provides additional gains (+2.6 AP) for instance segmentation. While reading work zone signs, composing a detector and text spotter via crop-scaling improves performance +14.2% 1-NED). Composing work zone detections to provide context further reduces hallucinations (+3.9 SPICE) in VLMs. We predict navigational goals and compute drivable paths from work zone videos. Incorporating road work semantics ensures 53.6% goals have angular error (AE) < 0.5 (+9.9 %) and 75.3% pathways have AE < 0.5 (+8.1 %).

cs.CV↗