Participatory Auditing for PRF Failures
_reachsumit · x · 2026-07-15
This paper discusses when PRF (pseudo-relevance feedback) fails and proposes a participatory auditing framework to identify which queries are negatively impacted by PRF.
The core idea is to first identify query types where "PRF worsens results" through participatory user research, then train an LLM reranker to learn these user preferences for selective query expansion.
The author emphasizes that this approach doesn't blindly apply expansion to all queries. Instead, it audits first to decide which queries should use PRF, thereby minimizing negative impacts on specific searches.
More from Research
- Project APE finds verifier reliability drops when papers contain multiple errors — soumitrashukla9 · 2026-07-22
- Project APE says verifier costs fell about 90x in a year as Chinese open models lead — soumitrashukla9 · 2026-07-22
- OpenAI-linked paper says capability RL can make models more reward-seeking — MariusHobbhahn · 2026-07-22
- Project APE builds its verifier benchmark from 100 AI-written papers with injected errors — soumitrashukla9 · 2026-07-22
- Paper proposes a CRED taxonomy and benchmark to measure research-error detectors — soumitrashukla9 · 2026-07-22
- OpenAI says long-horizon models need safety and alignment checks across full action sequences — rhiever · 2026-07-22