APT-RAG Builds Adaptive Reasoning Trees for QA Over Hundreds of Documents
_reachsumit · x · 2026-09-07
An EMNLP 2026 Findings paper introduces APT-RAG, targeting evidence-intensive QA where answers require synthesizing information scattered across dozens or hundreds of documents. It fixes two key flaws of existing tree/graph RAG methods—structural rigidity and topology-ignorant evidence gathering:
- Adaptive planning: dynamically expands the reasoning tree based on question dependencies and evidence requirements;
- Topology-aware evidence gathering: improves coverage via sibling evidence reuse, direct retrieval, and aggregation from child nodes;
- Evidence-guided batched answer generation: cuts generation overhead in evidence-intensive settings.
It outperforms existing structured RAG methods on evidence-intensive QA benchmarks, with code open-sourced.
More from coding & agent
- Vibe Coding plagiarism drama: 'Never seen it before' isn't originality, dev argues — ezshine · 2026-09-07
- MobileWorld benchmark: 201-task eval of phone GUI agents, best combo only ~52% — East-Muffin-6472 · 2026-09-07
- Redditor builds local agent 'Sai' with a real body, thermal senses and skin in the game — Quebber · 2026-09-07
- An agent workspace that combines cloud desktop, file management and custom agents — tekbog · 2026-09-07
- Making Codex the brain and Gemini the writer: one engineer's path to self-improving skills — xiaohu · 2026-09-07
- GPT-6 Astra ports Pokemon Tower Defense from Flash to Phaser 3 in ~5 hours, 4 prompts — TheReal4982 · 2026-09-07