Building Knowledge Graphs

This session explores how knowledge graphs support practical AI systems, from real-world use cases introduced in the Japanese edition of Building Knowledge Graphs to the role of graph databases in ensuring stable knowledge beyond changing LLMs.It highlights why RAG alone is insufficient and argues for keeping knowledge in durable, observable graph structures rather than inside models or embeddings.

Agenda:
* 5:40pm~6:00pm: Doors open, Checkin and networking
* 6:00pm~6:10pm: Implem
* 6:10pm~6:40pm: First Talk
* 6:40pm~7:10pm: Second Talk
* 7:10pm~7:40pm: Third Talk
* 7:40pm~8:00pm: Open discussion & Mixer, Doors Close

✨ What to expect:

• An overview of the Japanese edition of Building Knowledge Graphs and its purpose
• Practical knowledge graph use cases introduced in the book
• Why RAG alone is insufficient for stable, real-world LLM applications
• The importance of keeping knowledge outside LLMs and embeddings
• How graph databases provide a stable foundation as LLMs continue to change

Topics/Speakers:
🎤 Yuichiro Yasui, Senior Research Scientist at Nikkei Innovation Lab, Nikkei Inc.
Talk Topic: The Japanese edition of “Building Knowledge Graphs”
Talk Description: This book is the Japanese edition of O'Reilly’s "Building Knowledge Graphs," published in 2023.
Since there are so few Japanese resources on this topic, we published this edition to help engineers learn about knowledge graphs more easily.
In this talk, I will share some of the key use cases from the book.
Speaker Bio: To be added shortly

🎤 Sixing Huang, Research Scientist, Blogger and Neo4j Ninja
Talk Topic: Medical Intelligence with Gemini Enterprise Q&A
Talk Description: Information gaps in healthcare lead to real-world harm.
When patients lack the data their doctors have, they miss out on beneficial clinical trials and optimal treatment paths.
When regulators lack the data pharmaceutical companies have, we see public health tragedies like the OxyContin crisis.
To prevent these outcomes, we must democratize medical intelligence.
I’m here to show you how Gemini Enterprise Q&A changes the landscape.
We will explore how patients and professionals alike can use this tool to cut through the complexity of medical literature, find relevant clinical trials, and gain a transparent understanding of the drugs and diseases that affect their lives.
Speaker Bio: Studied in Bremen. German Bioinformatic Ph.D.

• Research Scientist, Blogger and Neo4j Ninja

About Gemini Data, Inc.

• A No-Code Graph Platform that redefines how graph data is used, understood, and shared
• Makes Graph Data More Accessible, Usable and Valuable to Everyone

🎤 Koji Annoura Neo4j Ninja, Graph Data & AI Enthusiast
Talk Topic: LLMs Change — So Where Should Knowledge Live?
Talk Description: RAG is often considered sufficient to make LLMs practical, but in reality,LLM outputs change significantly across models and versions. RAG provides information, but it does not guarantee stable interpretation or reliable decisions.
This talk discusses:

• Why RAG alone is not enough for practical LLM systems
• Why knowledge should not live inside LLMs or embeddings
• How graph databases enable stable, comparable, and observable knowledge

LLMs change. Knowledge should not.
Speaker Bio: Koji Annoura is a seasoned full-stack developer with over 40 years of experience in software engineering. Since 2009, he has been actively involved in Agile software development and co-founded the Neo4j Users Group Tokyo in 2013. In 2021, he also established the Apache Hop User Group Japan. Koji has helped numerous companies and teams in their Agile transformation, guiding them in adopting Agile and Scrum practices effectively. He is the author of A Technical Guidebook to Cloud Native Databases and The Practical Guide to MacOS X Server, and served as a technical reviewer for Graph Data Processing with Cypher.