Unverified50% confidenceFactExact time
LLM-IRA's final ranking is accomplished entirely by a single linear score fusion, requiring neither maintaining a full pairwise matrix nor online LLM calls to rerank item by item.
1
Sources
50%
Confidence
Long-term
Relevance
10/30/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedUsing inner product indexing on normalized vectors can often reuse efficient approximate index structures such as HNSW and IVF, without needing to build a separate index for cosine similarity.55% similarUnverifiedMeasures to deal with duplicate chunks include exact deduplication via content hashing, near-duplicate clustering based on embedding similarity, establishing normalization policies, and weighting by source metadata at query time, none of which require changing the model.54% similarUnverifiedTF-IDF+logistic regression never predicts the two rarest categories, and on the type field it is statistically on par with Gemma 4.54% similarUnverifiedLLM-IRA actively selects document pairs on which multiple retrievers' scores disagree most, asks the LLM to give a preference judgment, and then learns fusion weights for each retriever accordingly.54% similarUnverified格是由一组线性无关的基向量通过整数线性组合所张成的离散点集合53% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1114244API
curl https://kongchang.com/api/v1/knowledge/claims/1114244MCP
get_claim(id=1114244)