Reproduce “Qilin: A Multimodal Information Retrieval Dataset with APP-level User Sessions” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jia Chen, Qian Dong, Haitao Li, Xiaohui He, Yan Gao, Shaosheng Cao, Yi Wu, Ping Yang and 4 more
The Qilin dataset, from Xiaohongshu Inc. and Tsinghua University, offers real-world multimodal user session data, including interactions with Deep Query Answering (DQA) modules. This dataset facilitates research into integrated search and recommendation systems and Retrieval-Augmented Generation (RAG) in an authentic platform environment.
Key insights
- Deep Query Answering (DQA) modules reduce user browsing depth and clicks on organic results, suggesting a shift in user behavior when direct answers are provided.
- User engagement exhibits clear position bias (CTR decreases with rank) and session bias (engagement declines as sessions progress).
- Image-text notes generally achieve higher click-through rates than video notes across both search and recommendation contexts, potentially due to lower cognitive load.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.