March 2023 Summaries

Neo4j and Microsoft Azure have partnered to bring Neo4j AuraDS Enterprise, a fully managed graph data science cloud service, to the Early Access Program on Microsoft Azure. This allows data scientists to focus on analyzing data and getting ML models into production while infrastructure administration is taken care of. With over 65 pre-tuned graph algorithms and ML models, data scientists can easily identify meaningful relationships in their data and generate compelling visualizations. The partnership enables integration with Azure Machine Learning and Azure Synapse Analytics, opening up new possibilities to leverage the wider Azure ecosystem and create value quickly.

The Neo4j database uses a unique data type called "Path" to represent graph structures, which is different from traditional data types. The Path data type captures the structure of nodes and relationships as a result of a query. It can be obtained by assigning a pattern to an identifier like `MATCH p=`, and it has several functions available, including `length(path)`, `nodes(path)`, and `relationships(path)`. These functions allow users to work with Paths in various ways, such as getting the length or nodes/relationships of a Path. The Path data type also supports variable-length paths, where the user can specify patterns like `[*1..3]` to match paths with a minimum and maximum length. Additionally, there is support for path lengths 0, which allows users to perform operations on single nodes without requiring a full Path. However, it's essential to model the graph properly to avoid issues when using path lengths 0. The query engine will often return "every possible path," but in some cases, this can lead to traversal explosion, where the number of paths becomes too large, making it inefficient. To mitigate this, users can consider alternative approaches, such as using the `apoc.path.expand` procedure. Overall, understanding and working with Paths is crucial for effective graph querying in Neo4j.

Neo4j now supports AWS Managed Streaming for Kafka (MSK) through MSK Connect, configured as a source and sink with AuraDB Free. The process involves setting up an AWS account, creating an S3 bucket, installing Terraform on the laptop, and using Ed's Terraform module to build the environment for MSK and create sink and source connections between Aura and MSK's topics with the Neo4j Connector for Apache Kafka. The setup includes importing data into AuraDB using a CSV file of cities, their populations, locations, and counties, and then creating a graph that maps nodes with the City label to nodes with a Country label through the relationship LOCATED_IN.

This summary covers the key points of the text. The author, a head of product innovation and developer strategy at Neo4j, explores the output of `git log` as a graph in Neo4j AuraDB Free. He uses `jq`, a JSON processing tool, to parse the raw string by NULL bytes and output JSON for the fields. The data is then imported into Neo4j AuraDB Free, where it can be visualized and queried using Cypher statements. The author demonstrates various queries, including finding the most active authors, fixing parents with multiple commit hashes, and running long path queries. He also discusses profiling and optimizing Cypher queries for better performance. Throughout the session, he shares his experience, provides code examples, and offers tips for working with Neo4j AuraDB Free.

The newsletter covers various topics related to graph databases and their applications, including a new feature in Neo4j Bloom that allows interactive visualization of specific subgroups of nodes and relationships. It also mentions upcoming events such as Neo4j Live, Training Series, and GraphSummit Series, which include presentations on technical topics like AuraDB, routing web applications, and data governance. Additionally, it highlights the work of a speaker who presented a presentation on knowledge graphs and machine learning for halal food product recommendations. The newsletter also features a tutorial on building a full stack GraphQL application with Next.js, Neo4j AuraDB, and Vercel, as well as a tweet from Olaf Hartong about visualizing Microsoft Defender schema's in Neo4j.

Our latest 5-Minute Interview brings to you Arthur Zverko, an Israeli-based Software Team Lead at Equitybee, a company that helps startup employees exercise their stock options and receive pre-IPO shares from privately held companies. Arthur uses graphs in his work and advises graph newbies to learn Cypher from the beginning to build high-performing queries. He sees Neo4j as a natural choice for its developer-friendly approach and has seen surprising results, including two to three times faster query performance after switching to the database. Arthur believes that graph technology will continue to improve with the hosted cloud solution, and he is working closely with Neo to leverage the platform's capabilities and suggest improvements.

The text discusses the development of a JSON-based data model for Neo4j's property graph database. The goal is to provide a standardized way to describe and validate the schema of the database, allowing for better tooling and integrations with other applications. A proof-of-concept plugin has been developed to introspect the database and generate a JSON representation of the schema, which can be used for validation and visualization. The schema is based on a tokenized approach, where nodes and relationships are represented as classes and concrete instances, respectively. The algorithm uses existing Neo4j APIs to retrieve labels and relationship types, and groups them by sorted combinations to create node object instances and relationship instances. The JSON schema can be used to validate the data model and visualize its structure in a graphy visualization tool. The plugin is available for download and can be run within a Neo4j installation or built from source using Java 17. The authors invite feedback on the project, including comments on the JSON schema, data model materialization, introspector algorithm, and utility library.

Graph neural networks have shown great promise in learning representations of graph-structured data, which can improve the accuracy of downstream machine-learning tasks. The model presented in this blog post harnesses the power of graph neural networks to capture and encode the relationships between data points and enhance document classification accuracy. By leveraging word embeddings as input features, the GraphSAGE algorithm iteratively aggregates information from neighboring nodes, resulting in powerful node-level representations that can improve the accuracy of downstream machine learning models. The model's performance is enhanced by considering the relationships between articles, leading to improved precision and weighted precision compared to traditional word embedding models.

AuraDS Enterprise is a fully managed graph analytics and ML engine that allows data scientists to focus on analyzing data and getting ML models into production. This service takes care of all database administration, providing a library of over 65 pre-tuned graph algorithms and ML models to help identify meaningful relationships in data. Businesses can benefit from practical applications such as recommendation engines, anomaly detection, route optimization, customer 360, and network analysis. With AuraDS Enterprise on AWS, data science teams can integrate, analyze, and manage their pipelines using Amazon SageMaker and Amazon EMR, with a single SaaS license that covers all aspects of the infrastructure. As an AWS Data and Analytics Competency Partner, organizations can make the most of their data connections, influences, and relationships.

Neo4j Live is coming up next week, featuring a series of events including training sessions and meetups, as well as all Neo4j events such as webinars and live demos. The featured speaker for the event is Elena Kohlwey. Additionally, there are several topics being explored in articles and interviews, including knowledge graphs, natural language search, chatbots, and graph data modeling with Neo4j. Tomaz Bratanic discusses a knowledge graph-based approach to chatbots that provides explicit control over answers provided by the chatbot, while Gemini Data's use of GPT-3 in their natural language search feature allows users to query parts of a graph or get answers to specific questions. A chatbot ensemble for healthcare is also being explored, and an interview with David Meza from Neo4j at NASA discusses the company's work in this area. Finally, there are tutorials and guides available, such as installing Neo4j v5 on a Raspberry Pi 4 and using graphista to compute market data into a Neo4j graph database.

Neo4j Bloom is an advanced graph exploration tool that allows users to display different Neo4j graphs, explore graph data using the search feature, and expand it visually or by using custom Cypher queries. The new feature in Bloom, called the Slicer, enables users to interactively explore their graph data by changing values on nodes and relationships. This feature displays all distinct values of a property in ascending order, like a timeline or histogram, allowing users to easily slice the visual scene based on those values. Users can select specific slices of the set of total distinct values of a property using the Slicer chart bar, which represents the count of nodes and/or relationships that contain a distinct property value. The new feature is available in the Neo4j workspace and Bloom release 2.7.0. It allows users to create ranges by selecting properties and groups, and then select visible nodes and relationships based on those ranges. There are two types of range selection: manual, by resizing/dragging the range window on the histogram, and playback mode, which automatically changes the window in a specific period of time. The playback mode can be customized with three types of modes. The Slicer configuration is stored locally, and any change in the scene removes automatically the existing ranges.

Graph Neural Networks (GNNs) are gaining recognition in the machine learning community due to their potential for solving complex tasks, particularly in social networks, drug discovery, and recommendation systems. Unlike traditional neural networks, GNNs can operate on interconnected data with varying size, patterns, and complexity, allowing them to capture relationships between entities in a graph and perform tasks that require reasoning about network structure. GNNs are useful for dealing with complex systems that have interconnected components, making them a promising candidate for modeling many real-world problems. A blog series will explore the practical application of GNNs and related Graph ML, covering topics such as the promises and pitfalls of GNNs, Neo4j Graph Data Science, and alternatives to GNNs.

Angular application is built on top of the Nest.js backend framework with @neo4j/graphql integration, leveraging a NX monorepo for organization. The setup involves cloning a repository, installing dependencies, and setting up a .env file. A movie database is populated using Neo4j's provided dataset, and three pages are created: a list of all available movies, a movie details page, and a person details page. GraphQL queries are created to fetch data for these pages, including the `GET_MOVIES`, `GET_PERSON_DETAILS`, and `getPersonByName` methods. These queries utilize Neo4j's nodes and relationships to retrieve data from the database. The application demonstrates how quickly one can start building an entire app with @neo4j/graphql on the backend.

From Nodes to Rows: A Guide to Querying Neo4j Graph Database in Pandas/SQL Style Using Cypher` is a guide that explains how to perform data analysis using pandas and SQL, but instead of working directly with the relational database, it uses Cypher queries on a graph database like Neo4j. The author highlights the benefits of using Cypher for querying graph data, such as easier joins and flexible pattern matching. While some queries may be more verbose in Cypher compared to their equivalent pandas or SQL code, many times they are indeed simpler. The guide provides examples of how to perform common analysis operations like filtering, grouping, and aggregating data, and demonstrates the use of Cypher for tasks such as concatenating data and performing subqueries. Overall, the guide aims to show that Cypher queries can be used to analyze graph data in a pandas or SQL style, making it easier for users to work with their data without sacrificing the benefits of using a graph database.

The United Nations Sustainable Development Goals (SDGs) are a set of 17 interconnected goals adopted in 2015 to end poverty, protect the planet, and ensure peace and prosperity by 2030. The SDGs cover social, economic, and environmental issues, including no poverty, zero hunger, good health and well-being, quality education, gender equality, clean water and sanitation, affordable and clean energy, decent work, economic growth, and more. To support these objectives, a knowledge graph called SustainGraph has been developed to provide insights and recommendations for improving sustainability performance. SustainGraph is an open-source project that utilizes various data sources and sustainability metrics to track information related to the achievement of SDGs at national and regional levels. The project aims to create a unified source of truth for information related to the SDGs, using graph databases and natural language processing and machine learning techniques. A demo of SustainGraph was presented during a livestream, showcasing its capabilities in visualizing data and providing insights into how different policies impact the SDGs. As of now, the knowledge graph consists of 5M nodes and 10M relationships, with plans to include more resources in the future.

The text highlights an upcoming series of technical workshops, including an introduction to Neo4j on March 15. The workshops cover various topics such as AuraDB, GraphSummit Series, and a simple movie recommender system using Python and Neo4j. Featured nodes include speaker recommendations and tutorials on critical data analysis and data import into Neo4j. Additionally, there are mentions of special interest projects and events, including a chatbot for healthcare and new steering for Neo4j drivers. The text also features quotes from experts and users, showcasing their enthusiasm for the workshops and the potential applications of graph technology in different domains.

The article aims to provide a step-by-step guide on using Neo4j graph database for data exploration and analysis in Enterprise Resource Planning (ERP) systems, specifically SAP. It highlights the challenges of extracting data from SAP and provides a comprehensive guide on how to load sample SAP entities into a Neo4j graph database. The author uses Neo4j AuraDB Professional, deployed from the Google Cloud Marketplace, as an example. Once the data is loaded, users can analyze and explore relationships between entities using Neo4j's built-in tools such as Neo4j Bloom for visualization and graph data science algorithms. By leveraging these features, businesses can gain valuable insights into their product lines, business partners, and locations to optimize supply chain management and provide better customer service. The guide also covers how to apply AI/ML algorithms in Neo4j Graph Data Science to solve complex use cases like fraud detection and analytics.

Graph technology has applications across industries, globally, and in various use cases. Michal Stefanak, a programmer at VOLKE, shared his experience with graph technology, particularly Neo4j, which he believes is the only database that should be used. He discussed his work on the PHP driver for Neo4j, managing employee rights and requirements, and his enthusiasm for graph technology's scalability and ease of use, which has improved his programming experience. Michal thinks graph technology will continue to shape the future of development, making it faster and more efficient.

This project utilizes a knowledge graph-based approach to build a chatbot that leverages the capabilities of GPT-3 and Neo4j. The chatbot is designed to interact with users, provide information, and facilitate exploration and analysis of news articles. By integrating a knowledge graph, the chatbot can provide accurate and complete answers without hallucinating, making it suitable for mission-critical tasks or situations where lives are at stake. The project demonstrates how GPT-3 can be used to generate Cypher statements based on user inputs, and how a Streamlit application can be used to create an intuitive user interface for the chatbot. The code is available on GitHub, providing a starting point for developers to build their own knowledge graph-based chatbots and make these technologies more accessible.

The book "Full Stack GraphQL Applications" is designed to help full stack developers leverage the benefits of GraphQL, including its database integrations with Neo4j and cloud services like Netlify, AWS Lambda, Auth0, and Neo4j AuraDB. The author highlights key takeaways from his book in a series of videos, which also summarizes and updates the main points by explaining what was learned during a livestream. Additionally, there are upcoming technical workshops, meetups, and events such as FOSDEM 2023, Neo4j Live, and NODES 2023, featuring speakers like Smita Padhy and covering topics like graph data science, knowledge graphs, and facial recognition. The Neo4j Data Importer has also released a new feature to load more data sources without pre-processing, while William Lyon shares insights on keyword disambiguation using transformers and clustering in a NODES session.

The demand for conversational AI is expanding beyond large-scale language models, with many enterprises struggling with information silos that hinder the sharing and surface of helpful business information. Knowledge graphs, such as those built using Neo4j Graph Data Science, have proven effective in building internal chatbots that can accelerate projects and timelines by providing contextual awareness, personalization, natural language understanding, and multimodal interactions. By capturing relationships between entities, building user profiles, leveraging NLP and graph algorithms, and representing different types of information, knowledge graphs enable chatbots to provide more personalized, relevant, and sophisticated responses, ultimately improving business outcomes.

Neo4j's design system, Needle, was introduced to address issues of UI/UX consistency across its products as the company scaled. The team created a prioritization matrix to determine which components to build first and operationalized design by establishing a formal design system team, guidelines website, and code library. This has allowed for scalability without losing quality or efficiency, enabling designers to focus on research, testing, and collaborative efforts, while maintaining collaboration between engineering and UX teams. The design system now includes design guidelines, templates, processes, interactive capabilities, and provides benefits such as improved velocity, consistency, accessibility, and a stronger visual identity across products.