February 2020 Summaries

The newsletter covers various topics related to graph data visualization, Neo4j, and developer marketing. Jan Žák is featured as a community member who has contributed to the Neo4j community by sharing his expertise on full-stack web development, geospatial applications, computer security, and graph visualization. The newsletter also announces the release of BloodHound 3.0, which includes new features such as support for Neo4j 4.0. Additionally, it highlights the Gene Regulation Graph Database platform developed by a team at Guangzhou Medical University, which allows users to visualize and explore interactions between DNA and RNA proteins. The newsletter also celebrates Pokemon Day with a blog post on modeling Pokemon based on their generation, abilities, types, and moves. Furthermore, it promotes upcoming events such as the Neo4j 4.0 webinar and the Developer campfire talks at GraphTour Europe.

Realogy Holdings, the world's largest real estate company, leverages Neo4j to gain a holistic understanding of its customer ecosystem, comprising agents, brokers, and end consumers. The graph technology helps identify relationships between agents and their networks, which are crucial in retaining employees from leaving the company. Realogy chose Neo4j due to its industry standard, ease of use, and community support. To effectively utilize Neo4j, it's essential to prioritize use cases based on their potential to serve multiple purposes, avoid overloading the graph with too much data, and recognize when to say no to excessive requests. As graph technology continues to evolve, its future holds immense promise in accelerating innovation through machine learning and artificial intelligence by leveraging features engineering and utilizing graph algorithms as a standard practice.

Querying dense, exponentially growing networks in tall, relational databases is often unintuitive and poorly performant. Former CEO of Coshx labs Michael Wytock discusses how to use labels to limit search space as well as how to leverage stored procedures for performance improvements. He explains what patents are, why citation relationships are valuable, and the trends in patent data, highlighting how highly connected and quickly growing this data is. Wytock also talks about how they explored this data in Neo4j, as well as the three stages of implementation they experimented with. The presentation aims to provide a business case study on performance improvements using labels and stored procedures for complex graph queries.

The author of this text, a self-proclaimed Pokémon enthusiast, has created a graph database in Neo4j to visualize and analyze their Pokémon collection. The graph consists of nodes representing different types of Pokémon data such as Generation, Ability, Type, Move, and Pokémon themselves, connected by relationships that describe the characteristics of each Pokémon. The author uses this graph to explore various aspects of their team, including weaknesses against certain Moves, resistance to specific Types, and the strongest Pokémon in Sword and Shield with Dark-type moves. They also discuss potential applications for their Pokégraph, such as designing teams for battles or analyzing Type weaknesses and strengths, but acknowledge that there is still much data to be added, particularly from online sources like Kaggle and GitHub.

With the release of Neo4j 4.0 comes a new feature called Neo4j Fabric, which allows issuing Cypher queries that target multiple Neo4j graph databases at once. This capability can be used for data federation and analysis across separate databases, horizontal scaling of data storage and processing, or different hybrid deployments. The operational principle of Neo4j Fabric is to store shards as separate and disjoint graphs, where relationships are modeled using proxy nodes and correlating id values. A sharded data model can improve query performance by reducing the number of "jumps" across shards in complex queries. The LDBC Social Network Benchmark dataset was used to demonstrate the differences between sharded and non-sharded configurations and to explore considerations for graph sharding. The results show that Neo4j Fabric achieves impressive performance gains for complex queries, both in read query latency and in total read query throughput, with a carefully designed manual sharding scheme.

The Neo4j blog features a mix of community highlights, conference content, and new releases. Featured community members include Sylvia Tran and Jane Liang, who are organizers of PyLadies LA and have been involved in organizing graph meetups with their community. The blog also includes a behind-the-scenes look at the monitoring tool Halin, used by David Allen in his NODES 2019 session on monitoring Neo4j systems. Additionally, there is analysis of conference data, a new release of a Clojure driver for Neo4j 4.0, and developer guides on Cypher subqueries, multi-db, and Fabric. The blog also announces the upcoming Global Graph Celebration Day, an event to celebrate the birth of Leonard Euler and graph theory.

This week, a video demonstrating the upgrade process from Neo4j 3.5 to 4.0 is available on the Neo4j YouTube channel, showcasing the ease and speed of the migration process. The new version offers features such as unlimited scale and development agility, making it suitable for various use cases. To learn more about the upgrades and capabilities of Neo4j 4.0, viewers can check out a 7-minute demonstration video featuring Partner Solution Architect David Allen. Additionally, subscribers to the Neo4j YouTube channel will receive regular updates with graph tech-related content.

Jesús Barrasa, director of Telecom Solutions with Neo4j, discusses ontologies and their relevance in the field of artificial intelligence (AI) and machine learning. He begins by explaining that AI is not limited to machine learning, but also encompasses knowledge representation and reasoning, which is where ontologies come into play. Ontology is a formal representation of knowledge in a domain model, with three key characteristics: it must be machine-readable, explicit, and shareable. Barrasa uses examples from the FIBO ontology and schema.org to illustrate how ontologies can be used to make data smarter and reusable. He also discusses the two main uses of ontologies in Neo4j: interoperability and inferencing. Interoperability involves exposing data according to a shared vocabulary, while inferencing involves using knowledge fragments to derive new facts from existing data. Barrasa demonstrates how to use NeoSemantics to expose data as RDF and run queries on it, and shows an example of how to run inferences using the financial extension of schema.org. He concludes by highlighting the importance of ontologies in building a knowledge graph and provides resources for further learning.

The week's videos cover various topics related to Neo4j and graph data science. Vlasta Kůs and Golven Leroy demonstrate social media monitoring using Neo4j and NLP techniques, while also showcasing the Graph Data Science Library preview release. Additionally, they discuss features extracted from social platforms using GraphAware’s NLP tools. The Graph Data Science Library is a 4.0 Treasure Map that provides resources for users to learn about new features in Neo4j 4.0. Users can find information on multi-database support, security features, reactive drivers, and Neo4j Fabric through Jennifer Reif's curated blog post. Arrows Hacks shares usage tips and tricks for the graph modeling tool, while David Allen provides a short video guide for upgrading from Neo4j 3.5 to 4.0. Max De Marzi discusses ideas for improving Neo4j performance, and Alicia Frame announces the preview release of the Graph Data Science Library. Vlad Batushkov continues his series on building a Neo4j-backed flight search application, creating an APOC custom procedure to query flights. Users can also learn how to enrich existing graphs with data from the Wikidata SPARQL API through a blog post by the author. The week's featured community member is Elena Williams, who shares her passion for graph databases and community building.

Michal Bachman, the founder of GraphAware, has been working with Neo4j for several years, helping customers implement it and predicting that machine learning and AI will play a major role in the future of graph technology. GraphAware started using Neo4j to perform natural language processing on unstructured text in 2014 and has since developed a product called Hume, which is a graph-powered insights engine turning data into knowledge. Michal's favorite aspect of Neo4j is its new features in version 4.0, particularly the multi-database feature and the ability to apply security on nodes and relationships. The graph landscape has evolved significantly since Michal started working with Neo4j, from being a library for Java developers to becoming an enterprise-ready product with a vibrant community. Michal enjoys working with people implementing Neo4j, seeing the transformation it brings, and finding unusual use cases such as curriculum authoring systems and rules engines. He advises those getting started with Neo4j to start using it, focus on data modeling, and experiment with natural language processing. The future of graph technology holds promise in machine learning, with graphs being the underpinning layer for AI applications, and Michal predicts that Neo4j will move towards baked-in algorithms and more graph-based machine learning.

Neo4j is being used to model a simple business process, where a requestor submits a request, an approver reviews it and either approves or rejects it. The data nomenclature involved includes request, request state, lifecycle events, metadata associated with these events, actors involved in the process, roles assigned to actors, and the process itself. A graph is built based on this nomenclature to answer various questions such as finding all requests in a certain state, checking if a specific request can be approved or rejected, and seeing what comments were provided when a request was rejected. The use of indexes and relationships allows for efficient querying of the data, with some queries returning results in just one database hit. As the business process becomes more complex, Neo4j's graph structure allows for easy modification by adding new nodes and relationships without affecting existing ones.

Using a property graph model to surface relevant content is now common practice in many digital experiences. At Nordstrom, a one-step Markov chain implementation provided personalized homepage content on the mobile web experience, but scaling and iterating was challenging with relational data structures. To expand upon this success, a Neo4j graph database was built with website clickstream data, including product view and purchase data connected by shopper interactions for adult men's shoes. The team presented a simple initial concept for their graph that took into consideration just two steps of the customer journey: the current style being viewed and the most recent style viewed before that. They found all the paths of shoppers who had done that before and moved on to another item, suggesting the top item based on the number of paths observed. The team used Cypher queries to implement a "Viewed Next" feature, taking into consideration more context and returning the most popular styles. Recommendations are now a key navigation and discoverability tool for online shoppers, providing the right path forward and making a huge impact on customer experience and shopper outcomes. Graphs are well-suited for mapping a customer journey like this, allowing for efficient querying and aggregation. The team plans to continue investigating how to get graph-based strategies on par with their current system and apply this experience to other domains, such as modeling hand-curated outfits and creating tools for stylists and merchants.

</|im_end|>` Neo4j provides various tools for data import, including LOAD CSV and neo4j-admin import tool. Additionally, it can be connected to systems like ElasticSearch, SQL databases, MongoDB, and CouchBase using APOC procedures plugin. The Neo4j ecosystem is nearly complete for data manipulation. To expand this ecosystem, a Web Crawler can be used to obtain data directly from the web. A Web Crawler is a robotic program that specializes in browsing the web, extracting links, and storing content. Politeness rules must be respected when crawling websites to avoid overwhelming them. The Norconex Web Crawler provides an open-source tool for crawling and extracting data. Its structure allows for plugging in various collectors and connectors. A configuration file is used to connect the input (data collector) to the output (committer). Filters can be applied to extract specific data, and relationships can be defined between nodes. The California Grapes project uses Norconex to crawl a website about Californian wines, extracting grape varietals, regions, sub-regions, wineries, and other relevant information. The crawled data is stored in Neo4j, allowing for querying and analysis of the wine industry. Cleaning up the graph by removing unnecessary nodes and building relationships directly between subregions and wineries can improve the visualization. Querying the graph provides insights into related regions, sub-regions, wineries, and grape varietals. This project showcases the potential of combining Norconex with Neo4j for extracting linked data from web sources.

The Neo4j community has been active this week, with the official release of Neo4j 4.0 and various events and presentations, including a webinar on the new features by Dr Jim Webbers. The community has also shared videos from the NODES 2019 conference, FOSDEM Graph Room, and other sources, covering topics such as versioning graphs, schema-based access control in Neo4j 4.0, and Human Genes Graphs. Additionally, there are articles about various projects and applications built on top of Neo4j, including an HR People Analytics application and a single-page web application using Streamlit. The community has also recognized Vigneswaren Krishnamoorthy as the featured community member, who is a Neo4j Certified Developer and founder of a graph-based startup.

With the GA release of Neo4j 4.0, a new explainer video has been created to help users quickly understand its key features and capabilities in less than a minute. This latest version delivers significant improvements in scalability, security, agility, and reactive architecture on top of Neo4j's native graph foundation. The video is now available on the Neo4j YouTube channel, which also offers updated content weekly.

David Bader, a chair member at Georgia Institute of Technology's Computational Science and Engineering department, presented on predictive analysis using massive knowledge graphs. He discussed the implementation of graph algorithms in various domains such as social networks, transportation systems, storm evacuations, academic research, science, physics, chemistry, and astrophysics. The Spatio-Temporal Interaction Networks and Graphs (STING) group at Georgia Tech is working on solving real-world problems using graph analytics and surveillance. Bader aims to predict catastrophic events before they occur without experiencing similar tragic events first. He emphasized the importance of double-checking and reality checks when working with data and graphs. The presentation highlighted the potential of predictive graph analytics to anticipate past patterns of behavior and detect new threats, making it a crucial area for solving complex problems in various fields. Bader also discussed the use of Neo4j as a platform for building knowledge graphs and running graph queries to solve real-world problems.

Neo4j 4.0 is a major release that has taken over a year of work and has significantly improved the database's performance, scalability, and security. The new reactive data architecture allows for faster development, while the multi-database feature enables running multiple databases concurrently in a cluster or server, providing better isolation and flexibility. Additionally, Neo4j Fabric provides distributed queries across multiple databases, making it easier to process large datasets. The schema-based security model is also noteworthy, offering granular access control and preserving data integrity by considering the structure of the graph. This release is expected to make Neo4j more attractive to developers and users, providing a better experience for building fast, scalable, and secure applications.

The Graph Tour 2020 kicked off with events in Amsterdam and Mexico City, featuring presentations from the Developer Relations team and guest speakers like Amy Hodler. The event also announced Global Graph Celebration Day on April 15th. In other news, Neo4j Bloom version 1.2.0 was released with new features such as expandable legends and advanced styling options. The community shared various projects, including a QuickGraph of Itsu's food allergens and a blog post on exporting data from one Neo4j database to another. Additionally, there were updates on tools like Neo4j Query Log Analyzer and DB Analyzer, as well as new developments in the field of spatial data processing with GraphQL in the cloud.