KDD

AgentWorld: Personality-Aware Reliability Evaluation for Agentic Information Retrieval

AgentWorld：面向智能体信息检索的人格感知可靠性评估

Gunja Agarwal, Arup Kumar Das, Arun Menon, Jitesh Chandra Mishra, Vignesh Divakaran

机构 * PayPal（贝宝） 纠错

AI总结 AgentWorld框架整合多模块，通过三类实验验证其可评估智能体信息检索的人格感知可靠性，能发现统一测试未暴露的故障模式及轨迹级脆弱性。

Comments Agent4IR @ KDD 2026

PinSieve: Production Selective VLM Serving and a Governed Memory Flywheel for Enterprise Content-Quality Triage

PinSieve：面向企业内容质量分类的生产级选择性VLM服务及受控内存飞轮

Chuqing Gao, Yuanfang Song, Jonathan Zhang, Yifan Wu, Vishwakarma Singh, Qinglong Zeng, Andrey Gusev

AI总结 本文提出PinSieve，即面向企业内容质量分类的选择性VLM服务智能体，结合受控内存飞轮维护机制，提升审核效率、降低成本并改善信号交付，具有任务可迁移性。

Comments Accepted at the KDD 2026 workshop "Enterprise AI Agents: From Prototypes to Production."