The Path Matters: Evaluating Small Language Models Beyond Answer Accuracy in KGQA

arXiv:2609.27669 · cs.CL, cs.AI · Submitted 2026-09-23 · Read on arXiv

cs.CL, cs.AI

Submitted: 2026-09-23

Updated: 2026-09-23

Comments: 5 pages. Official implementation available at https://github.com/HalcyonSolutions/LLM_KGQA

Code: https://github.com/HalcyonSolutions/LLM_KGQA

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Fidelity — 21× in this paper · explained in 2 episodes
• Accuracy — 18× in this paper · explained in 1 episode
• Knowledge Graph — 7× in this paper · explained in 7 episodes
• Small Language Models — 7× in this paper
• Calibration — 5× in this paper · explained in 7 episodes
• Knowledge graphs — 4× in this paper · explained in 2 episodes
• Multi Answer — 4× in this paper
• Single Answer — 4× in this paper
• Pruning — 3× in this paper · explained in 4 episodes
• Adaptation — 2× in this paper · explained in 1 episode
• Path Edit Distance — 2× in this paper
• Scaffolding — 2× in this paper · explained in 3 episodes
• Ablation — 1× in this paper · explained in 1 episode
• Benchmark — 1× in this paper · explained in 2 episodes
• Frozen model — 1× in this paper · explained in 3 episodes
• National Yang Ming Chiao Tung University — 1× in this paper
• Reinforcement Learning — 1× in this paper · explained in 4 episodes

Sources

• Search-on-Graph-R1: Training Large Language Models to Search Knowledge Graphs with Reinforcement Learning
• Theseus in the Graph: Towards Traceable Multi-Hop Graph Navigation
• GraphWalk: Enabling Reasoning in Large Language Models through Tool-Based Graph Navigation
• Gemma 4 Technical Report
• The Llama 3 Herd of Models
• Ministral 3
• Olmo 3
• Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs
• Qwen2.5 Technical Report
• Qwen3 Technical Report

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering