EVENT DOSSIER
On September 7, 2026, the QoNext framework was officially proposed in the arXiv cs.CL preprint arXiv:2509.21889v3. This research aims to apply the principles of network and multimedia experience quality to the field of human-machine interaction evaluation in foundational models. The QoNext framework identifies key factors affecting user experience and collects human ratings using diverse configurations in simulated interaction scenarios, thereby constructing the QoNext database. The QoNext Model trained on this data serves as a neural predictor that can directly estimate user experience from measurable system parameters. The research results indicate that this framework can effectively decode the underlying mechanisms of user satisfaction and achieve precise human emotional prediction across various service conditions.
arXiv:2509.21889v3 提出 QoNext，首个将网络与多媒体体验质量原则适配至基础模型人机交互评估的框架。该框架识别影响用户体验的体验因素，并在模拟交互场景中通过多样化配置收集人类评分以构建 QoNext 数据库。基于此训练出的 QoNext Model 为神经预测器，可直接从可测量系统参数估算用户体验。研究结果表明，QoNext 能有效解码用户满意度底层机制，并实现跨多种服务条件的精准人类情感预测。