Lune

EMNLP Findings2025

GRADE: Generating multi-hop QA and fine-gRAined Difficulty matrix for RAG Evaluation

Jeongsoo Lee, Daeyong Kwon, Kyohoon Jin

Abstract

Retrieval-Augmented Generation (RAG) systems are widely adopted in knowledgeintensive NLP tasks, but current evaluations often overlook the structural complexity and multi-step reasoning required in real-world scenarios. These benchmarks overlook key factors such as the interaction between retrieval difficulty and reasoning depth. To address this gap, we propose GRADE, a novel evaluation framework that models task difficulty along two orthogonal dimensions: (1) reasoning depth, defined by the number of inference steps (hops), and (2) semantic distance between the query and its supporting evidence. We construct a synthetic multi-hop QA dataset from factual news articles by extracting knowledge graphs and augmenting them through semantic clustering to recover missing links, allowing us to generate diverse and difficulty-controlled queries. Central to our framework is a 2D difficulty matrix that combines generator-side and retriever-side difficulty. Extensive experiments show that error rates strongly correlate with our difficulty measures, validating their diagnostic utility. GRADE enables fine-grained analysis of RAG performance and provides a scalable foundation for evaluating and improving multihop reasoning in real-world applications.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

1 top-tier paper has cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers1

  1. A Temporal Reasoning Benchmarking Framework for LRMs via Difficulty-Controlled and Dynamic Test GenerationShide Zhou, Kailong Wang, Ling Shi, Haoyu WangISSTA 2026

Builds on5

  1. HopRAG: Multi-Hop Reasoning for Logic-Aware Retrieval-Augmented GenerationHao Liu, Zhengren Wang, Xi Chen, Zhiyu Li et al.ACL Findings 2025 · 24 citations
  2. Measuring Retrieval Complexity in Question Answering SystemsMatteo Gabburo, Nicolaas Paul Jedema, Siddhant Garg, Leonardo F. R. Ribeiro et al.ACL Findings 2024 · 6 citations
  3. Hop, Union, Generate: Explainable Multi-hop Reasoning without Rationale SupervisionWenting Zhao, Justin T. Chiu, Claire Cardie, Alexander M. RushEMNLP 2023 · 4 citations
  4. Do Question Answering Modeling Improvements Hold Across Benchmarks?Nelson F. Liu, Tony Lee, Robin Jia, Percy LiangACL 2023 · 1 citation
  5. DRAGIN: Dynamic Retrieval Augmented Generation based on the Real-time Information Needs of Large Language ModelsWeihang Su, Yichen Tang, Qingyao Ai, Zhijing Wu et al.ACL 2024

Related papers

  1. TRACE the Evidence: Constructing Knowledge-Grounded Reasoning Chains for Retrieval-Augmented GenerationJinyuan Fang, Zaiqiao Meng, Craig MacDonaldEMNLP Findings 2024 · 23 citations
  2. Cross-Granularity Hypergraph Retrieval-Augmented Generation for Multi-hop Question AnsweringChangjian Wang, Weihong Deng, Weili Guan, Quan Lu et al.AAAI 2026 · 2 citations
  3. When to use Graphs in RAG: A Comprehensive Analysis for Graph Retrieval-Augmented GenerationZhishang Xiang, Chuanjie Wu, Qinggang Zhang, Shengyuan Chen et al.ICLR 2026 · 56 citations
  4. REAL-MM-RAG: A Real-World Multi-Modal Retrieval BenchmarkNavve Wasserman, Roi Pony, Oshri Naparstek, Adi Raz Goldfarb et al.ACL 2025 · 33 citations
  5. MMhops-R1: Multimodal Multi-hop ReasoningTao Zhang, Ziqi Zhang, Zongyang Ma, Yuxin Chen et al.AAAI 2026
Dusk over the sea between two cliffs drawn in fine vertical lines