Reproduce “Result Diversification in Search and Recommendation: A Survey” 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

Haolun Wu, Yansen Zhang, Chen Ma, Fuyuan Lyu, Bowei He, Bhaskar Mitra, Xue Liu

This survey provides a comprehensive and systematically organized overview of result diversification in search and recommendation systems, introducing a unified taxonomy for diversity concerns, metrics, and technical approaches. It clarifies fragmented terminology and categorizes over 100 recent papers, offering a foundational resource for understanding and implementing diversity-aware retrieval systems.

Key insights

  • Diversity is a paramount criterion in retrieval systems, addressing issues like information overload, filter bubbles, and exposure unfairness for content providers, distinct from but often complementary to relevance.
  • Diversity concerns manifest differently in search (extrinsic for query ambiguity, intrinsic for redundancy) and recommendation (individual-level for user satisfaction, system-level for collective item exposure and fairness).
  • Effective diversity management requires a structured understanding of metrics (relevance-oblivious vs. relevance-aware) and approaches (offline methods like pre-processing, in-processing, post-processing; and online methods like bandit strategies and reinforcement learning).

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.