PatentsNET: Graph-Based Pretraining for Patent Analytics

Zhitao Yin, Yi Yang, Zhuoyi Peng, Zhenghan Zhang

Working paper. Major revision at a business journal Major Revision

A graph representation learning framework that models patent citation and ownership networks to jointly predict a patent's economic value and its litigation risk.

Federated Meta Embedding Concept Stock Recommendation

Zhuoyi Peng, Yi Yang, Liu Yang, Kai Chen

IEEE Transactions on Big Data (TBD) 2024

A federated meta-embedding framework for concept-stock recommendation that learns transferable representations across institutions while keeping sensitive financial data local.

[IEEE Xplore]

Agentic Graph Token Reasoning

Zhuoyi Peng, Yi Yang

Preprint. arXiv preprint 2026

An agentic framework that lets language models reason over graphs at the token level, interleaving retrieval and inference over graph structure.

[arXiv]

GraphInfer-Bench: Benchmarking LLM's Inference Capability on Graphs

Zhuoyi Peng, Jingzhou Jiang, Hanlin Gu, Lixin Fan, Yi Yang

Preprint. arXiv preprint 2026

A benchmark that systematically probes how well large language models perform inference over graph-structured data across a diverse suite of reasoning tasks.

[arXiv]

Beyond the Golden Teacher: Enhancing Graph Learning through LLM-GNN Co-teaching

Zhuoyi Peng, Hanlin Gu, Lixin Fan, Yi Yang

Preprint. arXiv preprint 2026

A co-teaching framework in which an LLM and a GNN supervise each other, moving beyond a single fixed teacher to improve graph representation learning.

[arXiv]