Self-Evolving Search Index
SELF-INDEX updates the retrieval keys behind an index by diagnosing misses, revising only the weak parts, and validating changes before they go live.
The paper frames search quality as a problem of index representations, not just retriever choice. Its system also uses a Query Simulator to test needs beyond the queries already observed. Across several corpora and retrievers, the authors report better retrieval than existing index optimization methods. They say the gains carry into search agents and memory systems for past interactions. HF Daily Papers' note
The paper frames search quality as a problem of index representations, not just retriever choice. Its system also uses a Query Simulator to test needs beyond the queries already observed. Across several corpora and retrievers, the authors report better retrieval than existing index optimization methods. They say the gains carry into search agents and memory systems for past interactions. HF Daily Papers' note
score 4