Code-on-Graph: Iterative Programmatic Reasoning via Large Language Models on Knowledge Graphs
The paper introduces Code-on-Graph (CoG), a new framework for integrating Large Language Models (LLMs) with Knowledge Graphs (KGs). CoG addresses the limitations of existing frameworks by providing a more flexible and scalable approach to programmatic reasoning. Experimental results show that CoG outperforms previous models by up to 10.5% on various benchmarks.
- ▪Knowledge Graphs are used to enhance the capabilities of Large Language Models.
- ▪Existing integration frameworks face issues of inflexibility and unscalability.
- ▪Code-on-Graph generates executable code based on KG schemas to improve reasoning.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2606.03705 (cs) [Submitted on 2 Jun 2026] Title:Code-on-Graph: Iterative Programmatic Reasoning via Large Language Models on Knowledge Graphs Authors:Weiwei Ding, Zixuan Li, Long Bai, Zhuo Chen, Kun Su, Fei Wang, Xiaolong Jin, Jin Zhang, Jiafeng Guo, Xueqi Cheng View a PDF of the paper titled Code-on-Graph: Iterative Programmatic Reasoning via Large Language Models on Knowledge Graphs, by Weiwei Ding and 8 other authors View PDF HTML (experimental) Abstract:Knowledge Graphs (KGs) are widely used to mitigate the limitations of Large Language Models (LLMs), such as outdated knowledge and hallucinations.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
