作者位次 #2/14SIGIR 2026Poster· Information Retrieval (Industry)
KARMA: Knowledge-Action Regularized Multimodal Alignment for Personalized Search at Taobao
Zhi Sun (Taobao & · Wenming Zhang (Taobao & · Wenming Zhang (Taobao & · Xingxian Liu (Taobao & · Yi Wei (Taobao & · Chenhe Dong (Taobao & · Xingxian Liu (Taobao & · Shengjie Sun (Taobao & · Zhixuan Zhang (Taobao & · Shengjie Sun (Taobao & · Chenhe Dong (Taobao & · Shengjie Sun (Taobao & · Xingxian Liu (Taobao & · Shengjie Sun (Taobao &
摘要
Large Language Models (LLMs) are equipped with profound semantic knowledge, making them a natural choice for injecting semantic generalization into personalized search systems. However, in practice we find that directly fine-tuning LLMs on industrial personalized tasks (e.g. next item prediction)...