Abstract: 

In our interconnected world, the broader context of data is crucial in influencing outcomes. However, current Business Analytics and decision-making practices often focus on isolated entities, neglecting the context. While Graph {Databases, Analytics, and Machine Learning} has been used for the last decade, only a few companies are able to leverage these Graph superpowers to their full extent. Large Language Models (LLMs), probably the most significant trend in the last year, are having a similar fate. Due to challenges such as hallucinations, internal knowledge access, and a lack of explanation of results, only a few companies can fully leverage the capabilities of LLMs.

Session Outline:

In this hands-on workshop, we will discover how the combination of Graph Data and LLMs can make each of them even more powerful. For this, we will cover the following topics:
 Graph Thinking, identifying, and modeling Graph problems 
 How to combine LLMs and Knowledge Graphs to: 
 Access custom data
 Minimize hallucination
 Provide explanations and traces of answers 
 Prompt engineering and fine-tuning to improve the LLM performance for specific use cases

Background Knowledge:

This workshop is hands-on, leveraging Jupyter notebooks, so there is no need to pre-install software and there are no prerequisites.

Bio: 

Jörg Schad is the ArangoDB CTO. Previously, he worked on Machine Learning Infrastructure in health care, distributed systems at Mesosphere, implemented distributed and in-memory databases, and conducted research in the Hadoop and Cloud area. He frequently speaks at meetups, international conferences, and lecture halls. Jörg is fluent in three languages and passionate about science & technology, education, and the environment.