September 2020 Summaries

The text discusses the Neo4j Online Developer Expo and Summit (NODES 2020), an event for graph professionals from around the globe. The author, a seasoned tech-community organizer, helped lead community partnerships for the event and was excited to involve friends and organizers from various tech groups. Over 30 community partners were onboarded, and their expectations were gathered through surveys. The participants expressed enthusiasm for talks on Graphs & NLP, CyberSecurity, SRE, and other graph-related topics, as well as live demos and networking opportunities. Many attendees highlighted the importance of graphs in complex systems and machine learning approaches, and some looked forward to sessions led by women tech leaders. The event aimed to foster collaboration and generate new ideas in the field of graph databases.

The Neo4j community has been busy with various news and releases. The FinCEN Files dataset, containing a vast network of industrial-scale money laundering, was published by the ICIJ and analyzed using Neo4j by Michael Hunger. In other news, Rik Van Bruggen explained exponential growth in a blog post, while Niels de Jong built a Spotify playlist generator using Graph Data Science Library. Ng Wai Foong navigated Singapore's transport network to build a subway journey planner. Additionally, there have been releases of NEuler, Graphlytic, and Neo4jClient .NET library. The community has also highlighted Max De Marzi as this week's featured community member, who is currently looking for new opportunities in graph databases.

The International Consortium of Investigative Journalists (ICIJ) has released a video exposing the global flow of over $2 trillion in dirty money, using graph technology to map out the FinCEN Files and reveal the failure of financial firms to stop illicit transactions. The investigation highlights the role of international banks in facilitating the movement of money from laundering, terrorism, trafficking, and other illicit sources.

Elsevier, a Dutch publishing company, is utilizing Neo4j graph technology to analyze its vast network of scientific content related to COVID-19. The project, Core Citation Graph (C-Graph), aims to query the network and provide insights into drug repurposing for COVID-19 treatment. By applying graph technology, Elsevier was able to build a polypharmacology model, predict effective drug combinations, and identify key proteins related to coronaviruses such as SARS-CoV-1 and MERS-CoV. The Neo4j platform enabled the team to collect and analyze data efficiently, producing simplified graphs that helped researchers understand COVID-19 mechanistic pathways in a matter of weeks. This technology is expected to be used beyond COVID-19 research, supporting studies into immunological diseases and enabling more targeted insights for researchers.

Chaos engineering allows companies like Google and Netflix to minimize downtime and surprise breakdowns in production by purposefully creating failures on a regular basis. The concept of chaos engineering has its roots in the idea of breaking things to explore toys as a child, and was popularized by Jesse Robbins, who founded GameDay to increase reliability by regularly creating major failures. Chaos engineering is now widely adopted by organizations, including Netflix, which promotes it online and provides business cases for its value. To conduct an experiment, one should define a steady state, utilize both a control and experimental group, inject failures into the experimental group, try to disprove the hypothesis that the system is resilient, and have a big red button to stop the experiment at any time. Tools such as Chaos Monkey, Mangle, and the Spring Boot Chaos Monkey tool can be used to conduct chaos engineering experiments. When it comes to Neo4j, monitoring known known elements and applications, and using tools such as Grafana and Failure Injection Testing (FIT) can help identify weaknesses in the system. Ultimately, chaos engineering with Neo4j involves simulating failures, injecting latency, and testing transaction faults, all while having a metric registry to monitor and control the experiment.

The International Consortium of Investigative Journalists (ICIJ) has exposed a vast network of industrial-scale money laundering through Western banks, with the help of Neo4j graph database technology. The FinCEN Files investigation reveals how money launderers move dirty money, involving drug cartels, corrupt regimes, and arms trafficking. Using Neo4j Bloom graph data visualization and exploration tool, journalists built a knowledge graph to explore over 400 spreadsheets containing data on 100,000 transactions, piecing together a nuanced picture of a broken system. The investigation draws from more than 2,100 suspicious activity reports, accounting for $2 trillion in dirty money. The FinCEN Files follow other breakthrough reporting and Pulitzer-Prize winning investigations by the ICIJ, such as the Panama Papers and Paradise Papers. The use of graph technology enabled the journalists to store, query, explore, and analyze the connections in their data, revealing dark secrets that were previously unknown.

The Neo4j community has been busy with various events and activities. A series of online events, Connections: Graphs in Government, took place last week, featuring presentations on building a smart city knowledge graph using Neo4j, Spring Data Neo4j 6, and other topics. The featured community member for this week is Matt Cloyd, a civic technologist who has worked on various projects related to government and technology. He will be leading an online meetup to introduce his new project Aspen, a markup language that produces rich graph data from plain text. Additionally, the Neo4j Developer Relations team is hiring two roles, including a Data Science Advocate and a DevOps Engineer. The community also received tutorials on building a restaurant recommendation engine using Neo4j and FastAPI, as well as getting started with Micronaut and Neo4j in Kotlin.

The Danish Business Authority uses graph data science to predict and prevent fraud, leveraging Neo4j for storing detailed data about every decision in a graph database for later review. The team has expanded their infrastructure to incorporate machine learning, event-driven architecture, and metadata strategy, providing an explanation traceability to their models. By analyzing the connections between companies and individuals in a graph, they can identify potential fraudsters more effectively than traditional methods. The project aims to improve explainability to ensure that decisions are transparent and understandable by humans, and the team is now looking into implementing this approach for future anti-fraud models. To introduce graph technology to management, it's recommended to start with a small part of data and demonstrate how it can connect any point in the data, making it easy to understand and deliver value.

Graph databases represent a significant departure from relational and NoSQL models, but this shift is inherently worthwhile. Graph Databases For Dummies, Neo4j Special Edition aims to introduce the basics of graph databases, providing practical advice for getting started quickly and confidently without delving into complex mathematics or coding. This book focuses on the fundamental building blocks of graph databases, including nodes, relationships, and properties, which support sophisticated, semantically rich queries at scale. Graphs are general-purpose data technology that can be used across various domains, offering benefits such as superior querying of complex models, ease of development, and flexibility in modeling data structures. The labeled property graph model is the most widely used graph database model, with specific naming conventions for nodes, relationships, and properties. Understanding who uses graph databases and why is crucial, as they are suitable for domains where data is connected, such as healthcare, finance, energy, or disaster response. Graphs bring several benefits across the system's life cycle, including high-performance querying, ease of development, and flexibility in modeling data structures. By embracing graph databases, developers can simplify their data models, reduce complexity, and create highly expressive and understandable data models that are highly performant.

Craig Taverner, a Senior Software Engineer at Neo4j, presented "Building Spatial Search Algorithms" at NODES 2019. The presentation focused on the two new data types introduced in Neo4j 3.4: temporal and spatial. The talk covered the history of Neo4j's work in spatial algorithms, including its recent internship with Eindhoven University student Stef van der Linde. Taverner discussed how the task given to van der Linde was to develop complex algorithms for spatial graph data, which led to some interesting conclusions about the performance of different data models. The presentation then delved into specific spatial algorithms, including route finding using Dijkstra's algorithm and A* search, point-in-polygon, polygon area calculation, intersection detection, distance calculation, and convex hull generation. Taverner also demonstrated how these algorithms can be used in a full-stack application to solve real-world problems. The talk highlighted the importance of speed and performance when it comes to choosing data models and writing complicated algorithms.

Neo4j is a graph database that enables money laundering detection by allowing for various types of queries to be viewed in different ways, including interactive dashboards with graphic visualizations and green, yellow, and red scoring of key performance indicators. The solution framework can ingest data from multiple sources, process billions of transactions per day, and detect suspicious activity in real-time and at scale. It also provides customizable plug-and-play components that allow AML staff to extend the base framework with money queries, custom graph data models, internal and third-party data sources and streams, ingestion APIs, and output APIs. By leveraging Neo4j's capabilities, compliance teams can better comply with AML requirements, reduce costs associated with fines and investigating false positives and false negatives, increase sales by improving brand-value reputation, meet the most stringent AML requirements for performance, availability, security, and agility at extreme scale, and get accurate predictions of suspicious activity.

This week's video features a session on getting started with Neo4j, Gatsby, and GraphQL. Will Lyon shows how to use a Neo4j sandbox backend with help from neo4j-graphql-js in his Weekly GRANDstack live stream. Adam Cowley builds a Knowledge Graph from Slack archives using APOC's Load JSON procedure, while Max De Marzi solves a tricky problem using the Traversal API. Meanwhile, Tomaz Bratanic dives into path finding algorithms in the Graph Data Science Library. The Neo4j community also celebrates Alexander Erdl, this week's featured community member, who has been working with the EMEA marketing team for six years and recently co-organized the Summer of Nodes event.

Carlie works with client teams in both the front and back office spaces to help them take advantage of new technologies, particularly knowledge graphs, which are enabling Accenture's clients to achieve significant impact. Knowledge graphs are 3D representations of data points connected by lines that provide context and meaningful storylines, allowing AI to make more useful decisions. This technology is opening up entirely new use cases for AI and radically shifting what's possible, enabling it to understand context and draw connections between data points like humans do naturally, which has the potential to be creative.

Graph algorithms are used to fight money laundering by increasing the accuracy of entity resolution, identifying high-risk payment chains and detecting networks potentially being used by high-risk accounts. The process involves creating pairwise-weighted relationships across a graph of similar entities using linear combinations of string and shared-attributes similarity scores, segmenting the graph into clusters with Weakly Connected Components algorithm, determining within each cluster which entity best represents others using Label Propagation graph algorithm, and utilizing stored procedures for Jaro-Wrinkler, Levenshtein and Sorensen-Dice algorithms along with graph attributes. Additionally, centrality algorithms detect liaisons, clustering algorithms identify subnetworks, and pathfinding and search algorithms identify payment chains and third parties layered between customers or transactions and other endpoints.

This week's Neo4j video features a variety of topics, including building a software knowledge graph using NLP and ontologies, faux bitmap indexes in Neo4j, k-means clustering with Neo4j, and connected feature extraction. The community also highlights the work of featured member Yusuf Baktir, who has been collaborating on the networkx-neo4j library. Additionally, Estelle Scifo's book "Hands-On Graph Analytics with Neo4j" provides a mix of hands-on exercises and explanations around NLP, spatial data, and graph algorithms. The community also shares updates from the recent Neo4j Connections: Knowledge Graphs event, including talks on building knowledge graphs using NLP and ontologies, faux bitmap indexes in Neo4j, k-means clustering with Neo4j, and connected feature extraction.

Graph data science is the approach Planethon takes to analyze massive datasets with vast knowledge about the Earth and its complexities. Planethon's mission is to bridge the gap between Earth science and business, helping companies build sustainable businesses by understanding how the Earth system works. The company connects large industrial players with science that informs them on sustainable investments, while also working with investors to make informed decisions. Planethon utilizes graph databases like Neo4j to analyze relationships within the Earth system and social systems, providing valuable insights and answering complex questions about sustainability and future investments. With a focus on R&D and technology development, Planethon aims to support companies in building sustainable businesses while amplifying its capacity for innovation and growth.

The Neo4j Labs extension plugin called Neosemantics (NSMNTX) helps users work with RDF and linked data by providing an interface to import, export, and manipulate RDF data in Neo4j. It enables users to import RDF data into Neo4j, which can then be queried using Cypher, and also allows for the export of RDF data from Neo4j as a lossless process. NSMNTX provides additional features such as ontology management, model mapping, inferencing, and cloning subgraphs, enabling users to create a more semantic and structured representation of their data in Neo4j. The extension is open-source and can be accessed through the Neo4j Labs page, where documentation, videos, and source code are available for contributors.