February 2021 Summaries

The Neo4j and Cloud Connections event has concluded with all the videos now available for those who missed it. Will Lyon showcased Chakra UI in the GRANDstack podcast app, while Jesús Barrasa explored the English WordNet graph. Tomaz Bratanic created an information extraction pipeline using spaCy, Wikifier, and OpenNRE to extract subject-predicate-object triples from unstructured text and import them into Neo4j. Meanwhile, Lju Lazarevic announced the Call for Papers for NODES 2021, which will be held on June 17th, 2021. Dr. Alessandro Negro released version 2.7.0 of Hume, GraphAware's graph-powered insights engine, with improvements to search and visualization capabilities.

The founders of Beijing We-Yun Data Co., Richard Wu and George Zhang, discuss the power of graph databases to store and correlate massive amounts of data easily. They share their experience with Neo4j, a graph database technology that helped them solve complex problems in finance, manufacturing, Chinese medicine, museum collection management, and even fighting COVID-19. The company chose Neo4j due to its reliability, ease of use, performance, and complete ecosystem. With Neo4j, they were able to close deals with big names in China, help clients with modeling and consulting work, and bring graph technology to many users in China. They believe that graph technology will continue to grow, especially in the areas of machine learning and AI, and are excited to work with Neo4j on this front. The company's goal is to expand their services to smaller companies where they can apply a different approach.

This project aims to explore artistic influence, cultural flows of concepts of sexuality and gender, and the formation of artistic communities through processes of migration and urbanization. The researchers have created a mock-up for grant applications, which will eventually be public-facing, allowing the public to look at queer histories, folks, and communities. They used Neo4j to explore histories in terms of letters, translations, and various flows through which this information and art traveled the world. The project's initial stages involved identifying figures, locating materials, and structuring data labels and properties. The researchers hope to adapt this project into a tool for public exploration on a large scale and receive funding as they scale up. They also aim to consider bigger questions like how race and gender are segregated within these communities and quantify the centrality of certain figures through token recognition.

Graph databases are an ideal fit for network and IT operations management, as networks are inherently graphs. Companies can store configuration information in real-time to alert operators to potential design flaws and reduce problem analysis and resolution times. Graph technology enables IT managers to catalog assets, visualize deployment, and identify dependencies between elements, allowing them to conduct sophisticated impact analyses and answer questions like which parts of the network depend on particular customers or which applications will be affected if a particular element fails. Comcast's use of graph technology in their smart home prototype demonstrates its potential for personalization and automation, as they developed a rich profile graph that models relationships between people, locations, devices, and personal information to provide a more personalized experience for users. Graph databases are designed to store interconnected data, making it easy to translate network and IT data into actionable insights, which is critical for effectively running an enterprise.

Hi everyone, Our video this week is an interview with NASA’s David Meza from Ashleigh Faith’s IsADataThing YouTube channel. Jesús Barrasa analyses the structured data of the White House website, Clair Sullivan imports data from Python, and I show how to use Speaker Listener Label LPA. Jack Collins creates Pokémon teams using Neo4j and the Graph Data Science Library. This week's featured community member is Tom Swinkels, a web developer at prepr.io who uses Neo4j for a headless CMS system. We discuss machine learning ethics and David Meza's experience with Neo4j at NASA, as well as various other projects and releases in the Neo4j ecosystem. My favourite tweet this week was by Mark Broadbent about current graph database status updates.

The Neo4j team has overhauled their Sandbox infrastructure, upgrading it to run on self-contained Docker images on AWS Fargate managed by a Lambda-Function-based API for easier scaling. The new Sandboxes now feature Neo4j 4.2 (Enterprise) with major updates including multi-database support, fine-grained permissions, and reactive drivers. The sandbox also comes with the latest versions of the APOC utility library, Graph Data Science Library, and Neosemantics RDF and linked data library. These features are used in their free online training classes, such as GraphAcademy. Users can now easily add their Sandbox as a Remote Database Connection to Neo4j Desktop and invite collaborators to share their Sandbox. The code examples have been updated with interesting queries for each dataset, and the datasets themselves are available on GitHub. The content of user Sandboxes will be available as dump-files that can be imported into Neo4j Aura (Cloud) and Neo4j Desktop. The team invites users to provide feedback through various channels.

The author discusses the concept of artificial intelligence (AI) and its various forms, including machine learning (ML). They argue that achieving AI is not just about building intelligent systems but also about understanding the structure of data, particularly graphs. Graphs are a powerful tool for analyzing complex relationships between entities and can be used to build intelligent systems without relying on ML frameworks. The author shares their personal history with graphs, from working on graph databases to applying graph search techniques to real-world problems like supermarket purchasing behavior and detecting spam. They highlight the benefits of using graph technology, including its ability to provide fast query results and extract valuable information from structured data. The author concludes that graph machine learning is a promising field that can help build intelligent applications without relying solely on ML pipelines.

In today's dynamic supply chain management, graph technology provides the agility and tools to generate real-time data-driven insights and predictive analytics, improving supply chain performance. By leveraging a graph database, companies can achieve global storage for a 360-degree view of their master data, enabling flexible performance and scalability, and making it easy to evolve as data sources change over time. This approach helps businesses navigate the complexities of transparency in their supply chains, with real-time decision-making capabilities no matter how large the dataset.

The Graph Data Science Library has released version 1.5, introducing supervised machine learning algorithms for node classification and link prediction. Meanwhile, Florent Biville shows how to build a Spring Data Neo4j 6 application on top of a graph of his Neo4j driver ecosystem metadata. Jesús Barrasa creates a graph of English WordNet using QuickGraph, while Alicia Frame and Amy Hodler announce the release of version 1.5. The community is also highlighted through features such as Gabriel Tard's role in improving Neo4j Graph Academy courses and Jennifer Reif's work integrating Neo4j with GraalVM.

Neo4j has become a crucial tool for Meredith Corporation's data analysis, enabling the company to better understand customer behavior by combining multiple streams of data into a unified identity graph. Prior to adopting Neo4j, Meredith Corporation struggled with traditional database approaches that couldn't effectively link and analyze their disparate data sets due to conflicts over timestamps and compatibility issues with cookies. The use of Neo4j's query language Cypher allowed the company to identify patterns in cookie behavior and create profiles of customers more accurately than before. Graph algorithms have enabled the company to scale their analysis, making it possible to process billions of cookies efficiently. The community support offered by Neo4j has also been instrumental in helping Meredith Corporation overcome challenges during development. Moving forward, the company plans to commercialize their project and continue to explore new ways to leverage Neo4j's capabilities.

The article describes a project that uses GraalVM with Neo4j to create a polyglot database, allowing developers to write custom procedures and functions in various programming languages. The project provides an example of how to use GraalVM to execute code in different languages, such as Java, Python, JavaScript, and Ruby, within a Neo4j database. The article covers the setup process for installing GraalVM and Neo4j, as well as the deployment of custom code as a database plugin. It also provides an example of how to use the polyglot procedures feature to execute code in different languages and retrieve results from the Neo4j database. The project can be used as a starting point for building custom procedures and functions that accept script files or write additional code in other languages.

The Neo4j Graph Data Science (GDS) Library has released its version 1.5, which brings several enhancements and new features, including graph embeddings, supervised machine learning, and easier production deployment. This release allows users to create an end-to-end model-building pipeline within Neo4j, enabling advanced ML techniques such as node classification and link prediction. Graph-native feature engineering is also introduced, allowing users to explore their data using graph algorithms and generate high-quality features for predictive models. The new release includes performance improvements and a low-memory format optimized for large graphs. Users can train supervised machine learning models, persist and share them, and manage them using the Neo4j model catalog. This update makes it possible for companies outside of Big Tech to take advantage of advanced graph-based ML techniques, marking a significant milestone in the field of graph data science.

The Under Armour Engineering Advocacy Team, led by Pat Brown and Karim Shehadeh, worked on a project called X-Ray to help non-engineers better understand the product journey. The solution used an X-Ray Graph that connects information from various data sources, including proprietary databases, Excel spreadsheets, and APIs, and pulls them into Neo4j, which establishes relationships in the product journey. This allowed for easy searching across the product catalog and aggregation of information. The team faced hurdles such as data intake, time, hiccups, and shortcuts, but learned valuable lessons along the way, including the importance of using indexes, avoiding relationship properties, and utilizing Cypher directives to simplify queries. The X-Ray project was built using GRANDstack, a powerful tool that enabled quick development and iteration. The team's work demonstrates the potential for Neo4j and GraphQL in product journey management and customer service applications.

In the context of contact tracing for COVID-19, graph visualization offers several advantages, including facilitating awareness of potential transmission occurrences without requiring data queries across multiple sources, adding date-based information to transmission chains to facilitate clinical decision-making, and enabling resource allocation based on chain statistics. The use of Neo4j, a graph database technology, allows for effective data modeling, access control, and browser interface development, making it an ideal solution for supporting healthcare professionals in tracing contacts and reducing transmission. A case study from the Geneva canton, Switzerland, demonstrates the benefits of graph visualization in tracing contact, highlighting the importance of accurate information on proximity in time and space to inform public health strategies.

Graph databases are being used by companies to solve complex problems, including anti-money laundering efforts. Traditional technologies struggle to connect the dots across multiple intermediate steps, requiring manual laborious effort from inspectors. Graph technology is better suited for this task, as it can recognize complex data relationships in real-time and accommodate new data sources without rewriting the data model. A money transfer service company used Neo4j graph database to detect "smurfing" activity, which thrives on splitting large sums of illicit funds into a hidden network of beneficiaries, allowing them to pursue criminal cases 20 times faster than with traditional tools. The use of graph technology provides a powerful weapon against the murky world of money laundering and embezzlement, enabling companies to improve detection and investigatory processes.

This week's video from the Build a GRANDstack Podcast App live stream covers various topics, including Next.js and GraphQL authentication, building a news knowledge graph using Natural Language Processing and Neo4j, and creating a tool called NeoDash that allows users to build Neo4j-backed dashboards without writing front-end code. Additionally, there are updates on the py2neo library, which now supports Python 3.9 and has improved performance, as well as new blog posts from Clair Sullivan and Tomaz Bratanic, who share their experiences with importing CSV files into Neo4j and creating a news monitoring data pipeline, respectively. The featured community member of the week is Jesús Barrasa, who has been part of the Neo4j community since 2015 and is known for his work on neosemantics, a plugin that enables the use of RDF and its associated vocabularies in Neo4j.

In a recent interview with Neo4j's Co-Founder and CEO Emil Eifrem, graph database technology played a crucial role in uncovering the FinCEN Files, a massive money-laundering scandal. Even with access to vast amounts of data, it requires specialized technology to identify patterns and connections that reveal the bigger picture. The interview highlights the importance of using the right tools to analyze complex data sets and uncover meaningful insights. Neo4j's graph database technology was instrumental in this case, enabling users to see the extensive web of relationships between financial transactions and individuals involved in the scandal.

Gram is a data graph format that makes it easy to create graphs as simple as `(a)–>(b)<--(c)`. It integrates with d3-gram, which allows for the rendering of graphs to SVG anywhere. To demonstrate Gram's capabilities, a notebook was created on ObservableHQ, showcasing how to import and manipulate Gram patterns using the `graphOf` function. The visualization can be modified by adding parameters such as zoom level and viewbox height, and it can also be used like a slider or text input field. By editing the Gram string passed into the `graphOf` function, users can create their own graphs, add text fields, load from files or URLs, and more. The example notebook re-created Wikipedia's Gallery of Named Graphs for fun.

Neo4j is being used with GraalVM to connect to a database from various languages such as Python, Ruby, R, Javascript, and others. This allows developers to use libraries in the source language for connecting to Neo4j from target languages like Python, or access non-Java libraries within programs in other languages. Polyglot procedures enable extending Neo4j and the Cypher query language by writing procedures and functions in any language and packaging them as a Neo4j database plugin. Additionally, polyglot Cypher allows using Cypher as the query language in various programs, allowing embedding Cypher code in Python or Javascript for executing against Neo4j. The tutorial covers setting up GraalVM and Neo4j, connecting to Neo4j from languages, and running operations on the database using Java, Python, Ruby, R, and other languages.

Making Sense of News, the Knowledge Graph Way` is a blog post that explains how to create a news monitoring data pipeline that combines natural language processing (NLP) and knowledge graph technologies using Neo4j. The data pipeline consists of three parts: scraping articles from an internet provider of news, running the articles through an NLP pipeline and storing results in the form of a knowledge graph, and enriching the knowledge graph with information from the WikiData API. The authors demonstrate how to perform simple network analysis and find insights using the knowledge graph. They also explore graph data science techniques such as co-occurrence networks, community detection, and centrality. The post provides a comprehensive overview of the benefits of using a knowledge graph to store news data and offers practical advice for implementing a similar data pipeline.

Graph databases are being used by companies to solve complex problems, particularly in the context of knowledge graphs. Traditional keyword-based search methods can be limiting and inflexible, whereas graph technology allows for more precise and meaningful information to be retrieved. A key example is NASA's use of Neo4j to connect data relationships between lessons learned over 50 years of space exploration, resulting in cost savings and improved project outcomes. By utilizing a knowledge graph, companies can improve access to information and provide users with the most relevant search results, leading to increased engagement and satisfaction.