Communicating Complex Results to Customers Using Graph Data Science… When You Don't Have a Front-End Developer on Your Team

Speakers:

• Dr. Dave Rench McCauley, Senior Director, Data Science and Machine Learning, Ernst & Young - Quantitative Scientific Solutions
• Dr. Michael Smith, Lead Scientist, Quantitative Scientific Solutions

Session type: Full Length

Abstract: In technical consulting, we help clients by building them cutting-edge tools to solve seemingly intractable problems. However, setting up infrastructure and building models is only part of the solution when it comes to providing value to customers. Visualization is the last-mile element that can make a client happy or leave them dissatisfied. In this talk, we will cover how we used Neo4j, Graph Data Science, advanced natural language processing, and the new Neodash visualization tool to track flows of scientific knowledge over time for our client, the U.S. National Science Foundation. We’ll discuss how we used language embeddings and automated clustering optimization to model the language of scientific articles (the “content”) and their positions within a citation network (the “context”) to model the evolution of ideas over time, at a scale that would be impossible with other tools. We will also cover how our collaboration with Neo4j, and specifically with the developer of the Neodash tool, enabled us to add the final ingredient of a successful project for this client: a front-end that allowed us to tell them an interesting and useful story while also providing them with a tool to realize the utility for their own work.

Enter the Matrix: Synthesising a Logical Digital Twin in Neo4j

Speaker: Spencer Shiotani, Principal Cognitive Software Engineer, Northrop Grumman

Session type: Full Length

Abstract: Maximizing efficiency in facility planning is difficult because buildings are complex systems. Tracking components such as electrical systems, security, network and data, people, seats, doors, etc., can become an insurmountable task. We propose a novel strategy to grapple with this inherent complexity through the utilization of graph-based Digital Twins in Neo4j. Producing a Digital Twin involves many different data sources, including architectural diagrams, HR data, and LiDAR scans to capture 3D representations of building interiors. In this talk, we'll demonstrate how Northrop Grumman creates and implements these Digital Twins to orient new team members, optimize seating assignments, communicate changes to the employee base, track resource usage, coordinate large-scale moves, and analyze the delta between what was planned and what actually exists, thus making a historically underutilized resource more accessible and manageable.

Cybersecurity Automation with OSCAL and Neo4j

Speaker: Alexander Koderman, Senior Developer, SerNet, Inc.

Session type: Full Length

Abstract: State-sponsored and state-tolerated cyber attacks continue to rise. Governments and regulators also continue to respond. Companies are facing an increasing number of compliance requirements and controls. The result is that assessment cycles are becoming faster and control satisfaction needs to be verifiable with high granularity down to single control statements for individual systems or even system components. The U.S. National Institute of Standards and Technology (NIST) has developed OSCAL, a machine-readable language for cybersecurity control implementation and assessment. The next step is to develop implementations to aid cybersecurity practitioners in their daily tasks, such as: determining control prerequisites, finding related controls, tailoring controls to the organization,and assessing control implementation. We demonstrate "OSCL4NEO4J" – a set of open source scripts and REST API that can be used to import and work with OSCAL data in Neo4j to solve practical problems faced by cybersecurity practitioners every day. The open source project that we present has already been recognized by NIST.gov and is referenced from their official OSCAL repository.

Neo4j Drivers Best Practices

Speaker: M. David Allen, Senior Director of Developer Relations, Neo4j

Session type: Full Length

Abstract: Neo4j & Neo4j AuraDB support Python, JavaScript, Java, Go, and .NET. In this session, we’ll cover some best practices for using Neo4j drivers in your application. We'll provide worked code examples of the most common things people try to do, and good patterns that will making coding with Neo4j easier. We will also dive into how querying a Neo4j database works, how Neo4j clusters operate, how queries are sent through the system, and what's happening under the hood of how drivers work.

XRP Ledger Blockchain ETL with Neo4j

Speaker: Thomas Silkjaer, Head of Analytics and Compliance, XRP Ledger Foundation

Session type: Full Length

Abstract: A story of ups, downs, learnings, and findings from working with representing the XRP Ledger in Neo4j. What started as a hobby project in 2018 to represent and analyze payments only, expanded in 2019 into a full history graph representation of the XRP Ledger blockchain that has been running for more than nine years with 2.2 billion transactions, generating more than 1.5 billion ledger objects. In 2022, the data model is updated to better scale with increased XRP Ledger use, reduce the storage footprint more than 50 percent by applying learning from the past years to remove redundant properties, move unused data to JSON strings that can be parsed with APOC as needed, and to reflect new possibilities with Neo4j 4.x. This talk also showcases how the database, that is kept in sync with the XRP Ledger +/- 10 seconds, is used in the fight against criminal finances by “following the money,” and how it is used to stay ahead of money laundering when criminals move funds quickly around prior to moving it to legitimate exchanges.

Analyzing a 55B-Entity Graph: Combinatorial Complexity and Business Decisions

Speaker: Dr. Janez Ales, Senior Research Scientist, Mathematician, Data & Algorithms, Knowledge Architecture & Innovation, BASF

Session type: Full Length

Abstract: World's Journal and Patent Knowledge Graph on 55 billion entities offers endless opportunities for traversals at scale, with challenges in data size and traversal execution times. However, todays limits of available cloud hardware speeds can be reached with algorithms on much smaller data sets due to combinatorial complexity. We discuss combinatorial tasks and challenges hidden in many industry data sets and decision problems, and highlight differences and benefits of graph database direct access via Cypher tools vs. language APIs on various use cases.

Exploring the Patient Journey of a Chronic Disease by Using Graph Analysis

Speakers:

• Danai Eleni Aristeridou, Data Scientist, Pfizer, Inc.
• Anastasia Karatzia, Data Scientist, Pfizer, Inc.

Session type: Full Length

Abstract: Harnessing healthcare data and unraveling the steps until the diagnosis of a chronic disease can be a challenging task. Revealing the path to a chronic condition could potentially lead to the early diagnosis, support treatment adherence, and adverse events mitigation. Every patient follows a unique path until the diagnosis of the target disease. Electronic health records and claims contain information about diagnoses, prescriptions, procedures, hospital admissions, and so on. Graph theory could capture relationships between the various data records, facilitate the identification of common paths between patients, and point out key comorbidities. In this session, we present a framework for the mapping of the patient journey by applying graph theory. The target is to better understand the progression of a chronic disease and reveal concealed connections between diseases that may not be visible with traditional visualization and predictive modeling techniques. Finally, we present how this framework can be transformed into a tool to quickly provide insights and recommendations to the decision makers.

Pouring Coffee Into the Matrix: Building Java Applications on Neo4j

Speaker: Jennifer Reif, Developer Advocate, Neo4j

Session type: Full Length

Abstract: Many of us have built applications for traditional data structures (like relational database tables), but is it different for graph data stores? Do developers need to retool and relearn? In this session, we'll cover a brief introduction to graphs, walk you through writing a typical Java application with Spring, and connect it to Neo4j. From interacting with the graph data from the application to deploying to the cloud, you'll see the process from start to finish. You'll also learn how to tackle pitfalls and pick up tips along the way, as well as explore the ways we can build, deploy, and connect applications to the database. This will come alive through a live demo, as we see the results of our efforts. Come to this session to build your business applications for graph data!