May 2023 Summaries

At Neo4j, a company that pioneers graph technology, being recognized as a Global Top 100 Most Loved Workplace by Newsweek is a significant achievement. This recognition is a testament to the culture the company has built and its commitment to employees. The award was decided in collaboration with the Best Practice Institute and surveyed over 2 million employees from businesses of varying sizes. Neo4j's open-source and developer roots form the core of a community mindset where collaboration between employees, customers, and developers is inherent. The company's approach mandates bottom-up input, creativity, and transparency to ensure every employee is part of its journey to build and scale the company for its customers, community, and market. With this recognition, Neo4j celebrates its culture that values relationships, collaboration, and user success, and invites others to join its story by exploring its careers page.

Cypher is a query language used for data analysis in Neo4j, and it offers various features such as SUM, AVG, MIN/MAX, and COLLECT for aggregations, COUNT for counting, filtering, pattern matching, multiple clauses, and subqueries for complex queries. The NodeS 2023 Call for Papers is open until June 30, with a submission deadline of May 31 to be considered for a featured speaker spot on the website. Upcoming events include MeetUp: Shortest Path in Python on June 1, 2023, APAC Training Series: Intermediate Cypher and Data Modelling on June 1, 2023, All Neo4j Events featuring webinars, live demos, and more, as well as the GraphSummit Series to connect with graphs.

The text discusses a project at Neo4j aimed at harnessing large language models (LLMs) with graph database technology, specifically knowledge graphs. The goal is to enhance the accuracy, transparency, and predictability of LLM output by integrating it with graph databases. Two initial use cases are identified: developing a natural language interface for knowledge graphs, which simplifies data selection, querying, and processing, and creating knowledge graphs from unstructured data sources, such as text documents. The project aims to improve the interaction between users and their connected data using Neo4j and LLMs, with prototypes to be developed in the future. The project is openly documented on a GitHub repository, where updates will be posted by the development team.

Neo4j DevTools have been integrated into Neo4j Workspace, offering a more cohesive developer experience with improved features such as refreshed parameter support, saved Cypher, powered-up graph visualization, new plan view options, and quick history search. The updated tools also include file filtering in the import job feature and the ability to stop an import job on request. These improvements aim to enhance the overall user experience for Neo4j developers.

The author of the text is a back-end developer at CytoSMART who has created a chemical graph database using Neo4j. The database contains a merge of three datasets: PubChem, NCI60, and ChEMBL. It includes six types of nodes: compound, synonym, cell line, measurement, experiment, and relation. The database is designed to store information about compounds, their synonyms, cell lines, measurements, experiments, and relations between them. The author provides instructions on how to download the dataset and set it up in Neo4j, which can be done in under an hour with minimal waiting time. The process requires 120 GB of free memory and is recommended for users who want to create their own database.

deps.dev is an open-source package dependency database that provides information about software packages, including their dependencies and vulnerabilities. The service analyzes data from various package repositories, such as npm, PyPI, Maven, Cargo, NuGet, and Go, to build transitive dependency graphs. This allows for the identification of problems in any package that affects other software they depend on. deps.dev offers an API that can be used to access this data, providing information about packages, their dependencies, and versions. The service also includes a UI where users can visualize the data, including graph views and hierarchical layouts. deps.dev can be integrated into graph databases like Neo4j to analyze software dependencies and provide insights into package vulnerabilities and relationships.

The Mock Graph Data Generator is a tool that creates deeply interconnected mock datasets for testing new ideas. It starts with designing the web by creating a network or graph of concepts, relationships, and nodes. The tool can generate graphs from prompts using an OpenAI API key, and allows users to export JSON files of the model. These files can be uploaded into arrows.app, a lightweight UI tool that generates shareable images. Once notated with additional properties such as names, locations, and dates, the data can be used to create CSV files compatible with graph databases like Neo4j. The tool is designed for property graphs, which store information in nodes and relationships with associated properties. With its compatibility with Neo4j's Data Importer tool, users can import the generated data into a running instance of the database. The Mock Graph Data Generator also includes features such as coordinate-based layout options and graph-centric query languages like Cypher. By using this synthetic data, researchers and developers can test hypotheses and explore new ideas in a controlled environment before working with real-world data.

BioCypher is a unifying framework for biomedical knowledge graphs that aims to make it easier for biotech experts and researchers to create knowledge graphs from various sources. The framework combines resources, ontologies, and output to help users load data, map expertly curated information, and export the combined data in different formats. BioCypher's approach is designed to be flexible and robust, with features such as unit testing and verbose feedback on knowledge graph structure. The team behind BioCypher plans to extend workflows per use case, ontology reasoning, improve performance, and develop a more intuitive user interface. By making the creation of knowledge graphs easier and more efficient, BioCypher aims to help researchers achieve their goals and make progress in the field of biomedical research.

Streamline your API development with Neo4j GraphQL and StepZen. Developers can quickly create GraphQL APIs that leverage data from multiple sources, including databases like Neo4j. StepZen simplifies the process of building a custom server from scratch, while offering tools and features to create flexible, powerful APIs that leverage the capabilities of Neo4j. With StepZen, developers can integrate data from multiple sources and extend the data in their GraphQL API. By combining Neo4j and StepZen, developers can simplify the process of building a GraphQL API and focus on defining the data schema and queries needed for their application.

This week's newsletter highlights a tutorial by Michael Hunger on model creation with JSON data from social media platform Bluesky using graph algorithms such as Louvain for clusters and PageRank for size/importance. The code is available on GitHub. The NODES 2023 Call for Papers remains open until June 30, and the first speaker has been announced. Featured Node Speaker Joe Cobbs presents a generalizable graph model for nesting teams within leagues across multiple seasons. Additionally, Daria Chadwick introduces Process Tempo, a graph application platform that creates powerful data applications built on Neo4j with features like no-code approach, advanced automation, and efficient data modeling tools. Various upcoming events, including Graphversation Ep, Neo4j Live, Jump into Graph, MeetUp, and APAC Training Series, are highlighted as well as a tweet from Javier Cha about his participation in the virtual symposium organized by Zhaojin Zeng.

The tutorial covers building a routing web application using Neo4j graph database, OpenStreetMap data, and Leaflet.js for rendering the map. It explains how to work with OpenStreetMap data in Neo4j, including importing and querying it, as well as adding address data from OpenAddresses. The tutorial also demonstrates how to use path-finding algorithms like Dijkstra's algorithm and A* to calculate routes between addresses. Additionally, it shows how to power autocomplete search functionality using full-text search indexes in Neo4j and display the route on a Leaflet.js map. The application allows users to search for addresses or points of interest and view the shortest route to them.

This week's newsletter from Neo4j highlights various topics related to graph databases, including experimental models using ChatGPT and Cypher query generation, as well as tutorials on using Neo4j with AWS architectures. The newsletter also features a call for papers for the NODES 2023 conference, updates on featured speakers and networking events, and showcases innovative applications of graph databases in various domains. Additionally, it shares a tweet that hints at an upcoming blog post about Neo4j's Ninja program, which promises to reveal insights into its inner workings.

Neo4j has announced the general availability (GA) of its native-graph cloud database management system, AuraDB Enterprise, on the Microsoft Azure cloud platform after a rigorous quality and early access trial process. The GA marks a significant milestone for Neo4j as it completes AuraDB's availability across all major cloud providers. For Microsoft enterprises, architects, and developers, this means ready access to the best native graph cloud database on their chosen cloud platform, with benefits including security integration with Azure Active Directory, private connectivity through Azure PrivateLink, ease of deployment and maintenance, and easy integration with developer tools such as Browser and Bloom. The GA announcement also comes after Gartner included AuraDB for the first time in its Cloud DBMS Magic Quadrant, recognizing Neo4j's financial and technical advantage. With this release, organizations can now purchase AuraDB Enterprise on Azure and count it towards their budget, appearing on the same bill.

This week, Bluesky, a social media application that serves as a reference for the Authenticated Transfer Protocol (ATP), has gained significant attention. With over 2 million people on its waitlist and 65k active users, it's an interesting case study. The platform's early Twitter-like feel, with a focus on fun and jokes, has attracted celebrities, journalists, and tech folks. While Bluesky currently lacks scalable moderation and user protection mechanisms, the ATP protocol itself is quite promising, using distributed IDs (DID) to identify users and storing data in cryptographically signed repositories hosted by federated "personal data servers". The Bluesky-Dev-Discord community has already grown to over 1000 members, with clients available in various programming languages. The author of the blog post imports the interaction graph from Bluesky using Neo4j AuraDB Free, visualizes and queries it, and runs graph algorithms for clustering and sizing. The process involves loading data into a Neo4j instance, creating user nodes, adding relationships between users and posts, and running graph algorithms to style and position them based on attributes. The author also explores the thread visualization and finds shortest paths between users via interaction networks or post-based threads.

Native graph databases are designed to be highly optimized for storing and processing graph data, offering advantages in performance, scalability, and efficiency compared to non-native graph technologies. Native graph databases store and process data as a graph, allowing for efficient navigation of relationships and connections within the data. They use index-free adjacency, which enables direct referencing of adjacent nodes without the need for indexes, resulting in faster query times and improved scalability. Non-native graph databases, on the other hand, are built on top of non-graph technologies, such as relational or NoSQL databases, and can struggle to handle large, interconnected datasets efficiently. Understanding the differences between native and non-native graph technology is crucial when evaluating databases for specific use cases.

NODES 2023 is a 24-hour online conference on October 26th, where attendees can expect varied content on graph-based topics in machine learning and other applications. The event aims to provide opportunities for developers and data scientists to present their projects and learn something new. A comprehensive email course, "30 Days to Master Neo4j", is also available, offering a structured approach to learning Neo4j Aura. The conference features various events, including Neo4j Live and GraphSummit Melbourne, as well as sessions on graph technology, ChatGPT, streaming data, and more. Experts like Mike Morley will share their insights on utilizing graph technology in team recommendations, while Mick Vleeshouwer discusses the architecture of a private Q&A engine with ChatGPT/LLMs. The event also includes tutorials on integrating Neo4j with AWS Managed Streaming for Kafka (MSK) and exploring NASA FIRMS data with Neo4j Aura.

NODES is a 24-hour graph community gathering that will take place on October 26, 2023. It's the fifth year of this event, which brings together developers, data scientists, and data engineers to learn about integrating graph technologies into machine learning and development projects. The conference features a lineup of over 100 speakers from around the world, showcasing their implementations, tools, models, and more, with sessions covering topics such as building intelligent applications, graph-powered machine learning and AI, and visualization. The event offers knowledge and skill development opportunities for beginners and experts alike, including interactive content delivered by fellow developers and data scientists. It also features a collaborative community aspect, where speakers engage in live talks and Q&A sessions, and attendees can build their skills while connecting with others. The Call for Papers is open now through June 30, 2023, and the conference tracks include building intelligent applications, machine learning and AI, and visualization.

NODES 2023 is scheduled to take place in October 2023, offering a unique opportunity for thousands of developers, data scientists, and data engineers to gather and learn about integrating graph technologies into machine learning and development projects. The event will feature over 100 speakers from around the world sharing their implementations, tools, models, and experiences, with a focus on building intelligent applications, graph-powered machine learning and AI, and visualization. Attendees can expect 24-hour live sessions across multiple time zones, covering the latest topics such as large language models, graph neural networks, and more. The event also includes an opportunity for speakers to submit their talks, with various types of presentations available, including 30-minute talks, lightning talks, and 2-hour workshops. NODES aims to provide a collaborative community where attendees can share knowledge, build skills, and exchange ideas.

GPT-4 has a great potential to generate Cypher statements based on only the provided graph schema, leveraging its extensive knowledge of various datasets and graph models during its training. However, limitations include issues with multiple aggregations using different grouping keys, version two of the Graph Data Science library being beyond its knowledge cutoff date, occasional relationship direction mistakes, non-deterministic behavior, and sometimes providing explanations despite instructions to avoid them. Despite these limitations, GPT-4 can be a valuable tool for experimental setups or when examples of Cypher statements are provided, offering impressive generalization ability in some cases.