Yueheng Wang | Searcharxiv

arXiv2026

Intuition First or Reflection Before Judgment? The Impact of Evaluation Sequence on Consumer Ratings

As online reviews increasingly drive consumer decisions, the impact of review interface design on rating authenticity remains under-explored. This research investigates how evaluation sequence ("Rating-First" vs. "Review-First") influences consumer ratings through three experiments and a large-scale secondary data analysis. The results reveal a significant polarization effect: in high-quality service contexts, the "Rating-First" sequence (vs. "Review-First") increases overall ratings, whereas in low-quality contexts, it leads to significantly lower ratings. This mechanism is driven by a serial mediation path of affective heuristics and cognitive effort. Furthermore, product attributes moderate this effect, with hedonic products amplifying the rating extremity compared to utilitarian ones. Secondary data from Yelp and Letterboxd confirm these findings, showing that the "Rating-First" platform (Yelp) exhibits a polarized bimodal distribution, while the "Review-First" platform (Letterboxd) shows more concentrated ratings. In conclusion, this research reveals how evaluation sequence shapes consumer ratings through affective and cognitive paths from an information-processing perspective. These findings extend the theoretical understanding of the online review formation process and offer practical insights for platforms to optimize interface design and enhance rating authenticity and credibility.

cs.IR↗

arXiv2026

Fedcompass: Federated Clustered and Periodic Aggregation Framework for Hybrid Classical-Quantum Models

Federated learning enables collaborative model training across decentralized clients under privacy constraints. Quantum computing offers potential for alleviating computational and communication burdens in federated learning, yet hybrid classical-quantum federated learning remains susceptible to performance degradation under non-IID data. To address this,we propose FEDCOMPASS, a layered aggregation framework for hybrid classical-quantum federated learning. FEDCOMPASS employs spectral clustering to group clients by class distribution similarity and performs cluster-wise aggregation for classical feature extractors. For quantum parameters, it uses circular mean aggregation combined with adaptive optimization to ensure stable global updates. Experiments on three benchmark datasets show that FEDCOMPASS improves test accuracy by up to 10.22% and enhances convergence stability under non-IID settings, outperforming six strong federated learning baselines.

cs.LG↗