LN: Hu et al. (2024) — GraphAgent: Agentic Graph Language Assistant

Bibliographic Reference

Citation: Hu, Y., Yuan, Z., Huang, B., et al. (2024). GraphAgent: Agentic graph language assistant. arXiv:2412.17029. EMNLP 2025. https://arxiv.org/abs/2412.17029

Important Note

Overview

The bibliography entry “Wang et al. (2024). The impact of sentiment polarity on ticket resolution time. arXiv:2412.17029” is incorrect. The verified paper at arXiv:2412.17029 is GraphAgent, not a ticket sentiment paper. The “ticket resolution time” paper does not exist in any academic database.

(This paper is in the PUMA project knowledge PDFs as “GraphAgent_Agentic_Graph_Language_Assistant_2412.17029v1.pdf”)


Pass 1 — Bird’s Eye View (5 Cs)

CAssessment
CategorySystem proposal
ContextApplies agentic LLM to graph-structured data and knowledge graphs
CorrectnessEvaluated on graph QA benchmarks. EMNLP 2025 acceptance.
Contributions(1) Unified agent for graph-structured and text data; (2) Plan-then-execute framework for graph operations; (3) Better than specialised graph models on hybrid tasks
ClarityGood.

Relevance: ⭐⭐⭐

Relevant for PUMA Stage 4–5 where the project knowledge graph (issues, dependencies, sprint history) could be represented as a graph and queried agentically.

MOCs