TechTalk AI Sharing Season: OpenKG Enters Ant Group – Knowledge Graphs and Large Language Models Empower General AI

On November 28, Ant Group's Ant Tech Research Institute hosted the "TechTalk· AI Sharing Season" event titled "OpenKG Enters Ant – Knowledge Graph + Language Model Empower General AI," featuring around thirty experts from the Chinese Society of Chinese Information and OpenKG who discussed knowledge graph technologies and their synergy with large language models.

Ant Group considers knowledge graphs a core technology, applying them in marketing recommendation, security anti‑money‑laundering, fraud detection, corporate credit, and medical insurance, and has established a Knowledge Graph Joint Laboratory with Zhejiang University to tackle key challenges.

Prof. Li Juanzi, a distinguished professor from Tsinghua University, opened the event by emphasizing that while large models are rapidly advancing, they still have limitations and need to be combined with knowledge graphs for complementary strengths.

Shao Qing, Director of Ant's Academic Cooperation Department, highlighted the strategic importance of knowledge graphs within Ant's AI roadmap and expressed hope that events like TechTalk would foster deeper collaboration between academia and industry.

Prof. Qi Guilin from Southeast University presented "A New Generation Knowledge Service Platform Integrating Large Language Models and Knowledge Graphs," reviewing the history of both technologies, their advantages and drawbacks, and arguing that their integration will usher in a true era of large‑scale knowledge for AI.

Bai Shuo, Chief Scientist and Director of Hundsun's Research Institute, delivered a talk titled "Ontology‑Native Computing," introducing a framework that treats events as special entities and integrates computational capabilities for event reasoning and evolution.

Yang Cheng, Associate Professor at Beijing University of Posts and Telecommunications, explored "Graph Foundation Models," proposing a new paradigm for graph learning in the era of large models.

Following the public sessions, a closed‑door meeting chaired by Liang Lei, Ant's Knowledge Graph Lead, featured speeches from Ant CTO He Zhengyu, who underscored the extensive use of knowledge graphs across Ant's services and announced collaborations such as the SPG framework, the OpenKG‑Ant joint lab, and the open‑source OpenSPG engine.

Ant Vice President Wang Zhirong outlined the company's AI strategy, emphasizing a full‑stack generative model approach that combines large models, small models, and knowledge graphs to drive industry applications.

Chen Huajun, Professor at Zhejiang University and initiator of OpenKG, discussed the history, mission, and open resources of OpenKG, as well as the semantic programming framework SPG and ongoing cooperative projects to build open common‑sense world graphs.

Finally, Ant's senior technical directors Gu Jinjie and Zhang Zhiqiang shared insights on key challenges in industry‑scale large model practice and knowledge‑enhanced AI applications, illustrating examples such as medical large models and Alipay coupon distribution.