March 2020 Summaries

As cryptocurrency continues to play a significant role in the future of finance, understanding patterns and outliers in Bitcoin transactions is crucial for link analysts. Graph technologies, particularly Neo4j's highly scalable graph database, complement each other perfectly with powerful graph visualization toolkit technology, such as KeyLines. By visualizing Bitcoin data, analysts can identify distinct shapes and patterns, including unusual transaction patterns centered around specific addresses. Understanding the nature of Bitcoin transactions helps fraud analysts identify patterns more easily, while customizing nodes and using styling techniques provide a clearer layout. The tool can also reveal insights that'd be hard to detect in an unintelligible list of anonymous transactions, making it perfect for spotting patterns and anomalies in intuitive and interactive ways.

Hi graph gang, I'm back to take over the TWIN4j reins. Thanks to Jennifer, Will, and Lju for curating the weekly graph news in my absence. In this week’s video, Calin Constantinov shows us how to make graph patterns show themselves in various datasets. We had the launch of the Neo4j BI Connector, TasoX explores Windows Events, and Rik begins his journey into the world of supply chain management. Several COVID-19 related projects have been created, including a project submission from the WirvsVirusHackathon, a graph-based data visualization of cases in Singapore using Neo4j, and a travel log for COVID-19 compiled by Alex Law. The Neo4j BI Connector delivers direct access to Neo4j graph data from business intelligence tools, allowing users to write SQL queries that are translated to Cypher queries and executed against a Neo4j database. This week’s featured community member is Siraj Munir, an advocate for graphs who has organized events in Karachi, Pakistan, and submitted an awesome project for Graphs4Good GraphHack – Knowledge Graph to the Rescue.

Keith Hare has joined the Neo4j team to spearhead language standards efforts. He is a busy person, continuing in his roles as Convenor of ISO SQL and GQL standards committee, and President of JCC Consulting, Inc., where he focuses on high performance, high availability, mission critical database systems running on OpenVMS operating system. Hare got started in the US SQL standards process over 30 years ago, and has since published three editions of the SQL Standard, adding support for temporal, JSON, Row Pattern Recognition, Multidimensional Arrays, among other capabilities. He is working with Neo4j to develop a declarative property graph query language called GQL, which will adopt SQL specifications for data types, operators, etc., and aims to provide a complete declarative database language for defining, managing, querying, and updating property graphs. Hare believes that another language is needed because the current standardization efforts focus on enhancing SQL, but he thinks GQL can bring more power to users of existing SQL data. The development process involves detailed standards writing, time-consuming comments resolution, and a roadmap calling for an ISO Committee Draft ballot starting in February 2021, with a published international standard by the end of 2022. Hare joined Neo4j's LANGSTAR team to work on GQL development, learn new technologies, and advocate for property graph standards.

With the rapid spread of COVID-19, global economies and healthcare infrastructure systems are facing unprecedented challenges, with individuals on the front lines experiencing uncontrollable chaos and others seeing the world come to a standstill. Neo4j's community manager notes that their community is made up of passionate and kind individuals who want to make a difference, leading to the formalization of the Graphs4Good GraphHack initiative. Unlike traditional hackathons, this event focuses on unity, collaboration, and positive contribution, with 150 people registering in just under a week. The initiative aims to bring together developers, data scientists, researchers, and practitioners to work on COVID-19 response topics and projects that make a real impact. Participants can either lead projects or contribute to existing ones by following simple steps, which include creating a GitHub project, adding relevant information, and registering for the event.

The Neo4j BI Connector is a solution designed to address the complexity of new data sources by providing a simple, programmatic, and supported way to interact with Neo4j and popular tooling. It allows users to retrieve read-only results from a Neo4j graph database for use in business intelligence (BI) applications, such as Tableau. The connector reduces dependencies on power users and custom ETL jobs, standardizes business intelligence workflows and architecture, provides access to real-time data, and increases collaboration by making it easier to share data with other groups within the enterprise. It works by exposing a virtual relational schema that is built by the driver, allowing users to issue SQL queries against Neo4j nodes and relationships. The connector translates SQL queries into Cypher queries, which are executed against the database via a standard Bolt connection, and returns the results back to the client. It supports various data types, including compound data types, and has an in-memory SQL Engine that can be used to satisfy any standard SQL query. The connector is designed to work with Neo4j 3.5 and 4.x series, requires APOC library installation on the server, and is compatible with popular BI tools like Tableau.

Neo4j Graph Database is designed to provide business agility by embracing change and innovation, with features such as full multi-database capabilities allowing multiple databases to be run inside a single Neo4j cluster, offering operational efficiency, security, and agility for B2B SaaS multi-tenancy. This enables organizations to deploy graph solutions on-premise or in the cloud, depending on their requirements, with options including managed services and self-service deployment. The ability to support multiple databases with the same schema is critical for multitenant environments, while flexible deployment options allow organizations to manage Neo4j themselves or use Neo4j Cloud Managed Services. As the future of the database is the graph data model, Neo4j has invested time and expertise in solving difficult computer science problems inherent in scaling a graph database, positioning it as the world's leading graph database for handling complex enterprise use cases.

The Neo4j community has been busy, with various opportunities coming up for projects, events, and content. The GraphTour Europe is going digital, and there's a Graphs4Good Hackathon to encourage projects that make the world a better place. Featured community members Eric Solender and Nikita Wootten share their work on rewriting the Neo4j Go driver from scratch. Additionally, there are resources available for improving Cypher performance with the query log analyzer, hands-on Twitch streams, and tutorials on Cypher projection using Lord of the Rings data. The community is also highlighting various projects and initiatives, such as COVID-19 analysis and biomedical applications of Neo4j.

The interview with Jesús Barrasa highlights the flexibility of graph databases like Neo4j, which supports rapid evolution and innovation in the telecom industry. Jesús appreciates the flexibility of the data model, allowing for agile iteration without painful migrations. He is also amazed by the performance of native graphs like Neo4j, particularly in path traversals and local queries. As someone with a background in graph technologies, Jesús brings expertise to Neo4j, having worked on various projects and PhD research in semantic technologies. He notes that learning Neo4j requires understanding the engine's secret sauce: how data is stored, which enables building effective models and queries. In the telecom industry, graphs can have a significant impact, particularly in network management, service delivery, and consumer-focused solutions. Jesús believes Neo4j will continue to grow in importance as automation of services becomes more prevalent with 5G and the Internet of Things. Finally, he advises someone getting started with Neo4j to download the software, play with it, and seek expert advice when needed to ensure success.

Machine learning is a field of statistics that enables machines to learn with data, which is often used to automate tasks and make predictions. Lauren Shin, a developer relations intern at Neo4j, presented on machine learning and its applications in graph databases. She introduced three approaches to better data analysis through machine learning: Approach 1 uses user-defined procedures in Cypher to train models from the Neo4j browser while keeping data within the graph; Approach 2 eliminates data export by using graph databases to store and hold data during model development; and Approach 3 restricts comparisons with clustering, where graph structures are used to infer groups of similar users. Shin also discussed model version control, which represents relationships between different versions of models in a graph, increasing transparency in the process of building machine learning models. Additionally, she highlighted the potential of new graph-based machine learning algorithms that rely on graph-based input instead of vectors, offering insights and benefits not available with traditional vector-based methods.

Neo4j provides robust and granular security, allowing developers and administrators to control access to data at a fine-grained level. This is achieved through role-based access control, where users are assigned roles with specific permissions to read or write certain types of data. The security rules are defined using database and schema information, ensuring consistency across Neo4j databases, clusters, and shards. With Neo4j's identity and access control features, including Kerberos and LDAP, communications with the database take place over a secure binary protocol or HTTPS requests. The graph database simplifies the task of assigning permissions in a graph data structure, empowering security staff to create rules about sensitive data that are applied at the database level. This approach ensures modern applications have embedded security regardless of where data physically resides, making it an effective solution for governance and security critical data.

Hi Graph Friends, if you haven't already heard, GraphConnect 2020 in New York City has been postponed and the remainder of the GraphTour Europe events have been moved to online only. The Neo4j community is working with analyzing COVID-19 data in Neo4j, including queries for importing data from Johns Hopkins into Neo4j to model COVID-19 infections throughout the world. Additionally, a new Neo4j Sandbox project for Graph Data Science has been launched, showing how to use the Neo4j Graph Data Science library to run graph algorithms such as PageRank and community detection. The GRANDstack team published a 5-part blog series on access control in GraphQL APIs built with GRANDstack, exploring topics like schema directives, query transformation, and authorization rules. The Neo4j community is also working on new Visual Studio Code extension for Cypher and Neo4j, as well as a new Wikipedia page for the GQL graph query language. Stay safe out there and we'll see you next week!

This week, our Analytics and AI Manager Amy Hodler gave a talk on Responsible AI, emphasizing the importance of ethical AI development that aligns with social values. The presentation explores real-world cases and implications of irresponsible technology to create a responsible AI environment free from myths, bias, inappropriateness, and more. The talk is available on our Neo4j YouTube channel, which is updated weekly with various graph tech content.

To check for data inconsistencies in a Neo4j graph database, you can use Cypher to examine your nodes by node label and print out node labels having different property key sets. The process involves creating an initial Cypher query that works for a single Node Label, then using the returned Node Labels to create individual Cypher statements for each Node Label. These statements are then run using APOC's `apoc.cypher.run()` function, which returns results in a map format. The final step is to process these results to determine which Node Labels may have different property keys, and to check if there is more than one row from each Cypher query, indicating inconsistent data. By running this query, you can identify potential issues with your data model and take corrective action.

Risa Myers' talk at NODES 2019 discussed her experiences teaching Neo4j to students, highlighting the importance of selecting the right tool for the job and adapting to different database types. Tomaz Bratanic has released a blog post featuring migration guides from the Graph Algorithms library to its successor, the Graph Data Science (GDS) library, with examples from the Graph Algorithms book. Adam Cowley has joined the Developer Relations team, working on Neo4j Developer Tools, and shared various topics he's been exploring, including multi-database features in 4.0, building a lightweight control panel to automate database spinning up, and an early version of a Cypher Query builder for Node.js. Estelle Scifo has released a new version of neomap with routing, which combines well with the Graph Data Science library. The team also highlights various quick links, including COVID-19 visualization, composite indexes, Neo4j Simply in Depth, GraalVM for Python and other languages, and Auth Strategies for Drivers. A featured community member is Thomas Silkjær, a designer, developer, and data analyst passionate about blockchain technology and sharing his knowledge with others.

GraphConnect 2020 features two days of packed agenda with keynotes, presentations, and lightning talks on various topics such as Neo4j Tech Talk, graph data science journey, use cases, and best practices. The event includes keynote speakers like Emil Eifrem, CEO of Neo4j, and Albert-László Barabási, a renowned network specialist. The agenda also covers recently released products and solutions in the Neo4j suite, including Neo4j 4.0, Innovation Labs, AuraDB, and graph visualization tools. Attendees can participate in GraphClinics for free consulting and troubleshooting, DevZone with Neo4j engineers, and mingle at the disConnect Party to build relationships with fellow engineers and business execs.

Modern applications require intelligence and learning to support smart decision-making in real time. They must be able to scale without limits, meet unexpected demand, and provide straightforward security and flexible architecture. Graph databases are emerging as a solution for intelligent applications, offering a flexible and scalable foundation for connection-oriented applications. They use nodes and relationships to store data, enabling queries that can run in fractions of a second, and support both graph transactions and graph analytics. Graph data science unlocks insights from connected data, providing algorithms such as pathfinding, centrality, community detection, heuristic link prediction, and similarity. These technologies are crucial for building intelligent applications that meet the evolving requirements of modern software development.