Reproduce “A Survey of Large Language Models on Generative Graph Analytics: Query, Learning, and Applications” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Wenbo Shang, Xin Huang
Researchers at Hong Kong Baptist University conducted a survey of Large Language Models (LLMs) applied to generative graph analytics (LLM-GGA), providing a new taxonomy to categorize methods for graph querying, inference, learning, and various applications. The work compiles benchmark resources and identifies current challenges and future directions for leveraging LLMs to process and reason over graph structures.
Key insights
- LLMs can solve simple graph problems and some NP-hard problems on small graphs but generally lag traditional algorithms in efficiency and accuracy for complex or large graphs.
- Knowledge Graphs (KGs) effectively mitigate LLM limitations such as hallucination and poor factual recall, while LLMs enhance KG tasks like completion and construction.
- LLMs contribute to graph learning as enhancers for GNNs, direct predictors, and graph generators from text, highlighting versatile roles but also architectural integration challenges.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.