Solving Exclusion Inversion in Retrieval: EXCISE Boosts Success Rate to 69%
_reachsumit · x · 2026-08-07
Late-interaction retrievers like ColBERT handle exclusion queries poorly, suffering from "exclusion inversion" where excluded topics are ranked higher. The paper notes that the difficulty lies in identifying the excluded topic based solely on the query, making it impossible to recover constraints from frozen vectors.
The authors propose EXCISE, a query-time fix that corrects this inversion while leaving the index frozen. It uses two query-side modules totaling 1.5M parameters to identify the topic and re-embed a 100-document shortlist, alongside a parameter-free rule to demote matching candidates.
Across six collections and three backbones, EXCISE outperforms all baselines. It raises exclusion success@10 on ExcluIR from 0.058 to 0.691 and Boolean NOT accuracy from 0.25-0.29 to 0.90-0.92. The authors also release X-BENCH, a tiered benchmark for explicit, implicit, and compound exclusions.
More from Research
- Training LLMs Isn't Rolling a Ball Downhill: Loss Landscape Video Reveals the Mystery of Gradient Descent — emax · 2026-08-24
- LLM Preference Tuning Fails Under Domain Shift, Study Shows — nikaletras · 2026-08-24
- Paper: Agents struggle to evolve via explicit skill libraries — rohanpaul_ai · 2026-08-24
- Francis Crick Institute Launches AI for Biological Discovery Seminar Series — MihaelaVDS · 2026-08-24
- ETH Zürich trains robot to play badminton using whole-body reinforcement learning — lukas_m_ziegler · 2026-08-24
- ParaTempo: Efficient Parallel Reasoning via Temporal Confidence — SJTU · 2026-08-24