Negative Sampling Techniques in Information Retrieval: A Survey
2026-03-20T08:52:19Z•01b51da48e9d4e50b4908458a64e46a0c92b75fee223232975a086530bac219f
API-integrationLLMMLOpsRAGauditabilitydata-exfiltrationdataset-contaminationevaluation-benchmarkshealthcare-safetymodel-biasmodel-hallucinationpatient-safetyprivacysynthetic-datatool-invocation
What happened
This collection of recent IR/LLM papers highlights several security- and safety-relevant developments: increased use of retrieval-augmented generation (RAG) and dynamic API/tool invocation (DynaRAG) which expand the attack surface (data exfiltration, malicious tool calls, supply-chain and third-party API risks); deterministic evidence-gating and sufficiency classifiers that improve auditability and reduce hallucinations; LLM-driven synthetic negative sampling and dataset synthesis that raise risks of data contamination and privacy leakage; healthcare-specific models (HypeMed) that, if misappli
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_cs_ir
- Record identifier
- 01b51da48e9d4e50b4908458a64e46a0c92b75fee223232975a086530bac219f
- Enrichment time
- 2026-03-20T08:52:19Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.