November 2021 Summaries

It's an exciting time for graph technology as it opens doors to new efforts in data exploration and management, accelerating digital transformation. With innovative technologies like Neo4j's graph data platform, quantifying business value can be challenging due to the complexity of the capabilities it enables. A credible example of return on investment is necessary to support a business case, which is where Forrester Consulting's Total Economic Impact assessment comes in. This proven industry-standard framework evaluates four key pieces of the puzzle: cost, benefits, flexibility, and risk, taking into account associated impact on the business. The study found that Neo4j can pay for itself more than 4x in three years with a 417 percent ROI, and customers realized improved business results of 20 percent on average. Additionally, accelerated Time-to-Value was up to 60 percent, and digital transformation cost savings were achieved.

This week, the Neo4j community has been busy with various articles and projects. Adam welcomes readers to Thanksgiving and thanks them for their engagement despite the holiday period. Featured community member Jeremy Grignard is highlighted for his work on Pegasus, a knowledge graph supporting early drug discovery. Manning announced the early access release of "Graph Algorithms for Data Science" by Tomaž Bratanič, a must-read for fans of Tomaz's articles. Clair Sullivan published an article on making FastRP graph embeddings work for specific problems and released part 16 in her Bite-Sized Neo4j for Data Scientist series. GraphXR was used to explore fraud rings, while Jim Webber and Antoine Larmanjat co-authored an article on resolving financial services challenges with Neo4j Aura on Google Cloud. Nathan Smith explained how to create a similarity graph using properties of a Neo4j Graph. Andreas Berger presented new features in 1.3 and 1.4 of the Neo4j-GraphQL-Java library, while a brand new course was released on GraphAcademy for building Neo4j applications with Node.js. Christophe Willemsen shared his thoughts on graph visualizations, styling, and presenting information.

The dataset is a consumer complaints database from data.gov, containing over 2.3 million complaints about products from different companies. The Neo4j team re-enacts a popular webinar on loading CSV data into Neo4j using AuraDB Free, and imports the data into Neo4j to demonstrate its capabilities. The dataset includes nodes for Complaints, Companies, Products, SubProducts, Issues, SubIssues, Responses, and Tags, with relationships between them. The team demonstrates how to load and process the data, including filtering, merging, and querying the graph. They also provide examples of exploratory queries, such as finding sub-issues with communication tactics or typical response percentages per product. The dataset is available for exploration and analytics using Neo4j.

Neo4j GraphQL 2.4.0 has been released with a significant feature that simplifies data retrieval: autogenerated aggregation queries. These queries enable users to retrieve statistical information about nodes, relationships, and properties in their graph without having to write custom Cypher logic. The new feature allows for automatic generation of queries for top-level aggregations, where aggregations inside an existing `WHERE` clause, and field aggregations across relationships. This update provides a more streamlined experience for users, making it easier to answer complex questions about their data.

Shell companies are often used for illicit activities such as money laundering, sanction evasion, bribery, and financing terrorism due to their ease of incorporation, disposability, and anonymity. Researchers have developed a model using graph analytics in Neo4j to detect these entities, which can be identified by analyzing those already accused or identified in money laundering cases. The UK was chosen for its public register of ownership for companies, which revealed a dichotomy between theoretical standards and observable activity, providing an interesting case study. A dataset of 800 shell companies was used, with information collected from various sources including corporate databases and government registries. Graph database and graph data science were used to analyze the data, revealing valuable insights such as individual executives with multiple appointments in numerous entities, and a significant gap between theoretical standards and observable activity. The research also explored the use of graph algorithms and supervised learning to improve performance metrics, setting a base for future research work.

The Neo4j community has been actively sharing their knowledge and experiences through various articles, presentations, and videos. This week, the focus was on the Neo4j Health Care & Life Sciences Workshop 2021, which featured in-depth technical presentations from the healthcare and life sciences space. Will Lyon demonstrated how to build location-aware personalized news recommendations using Cloudflare Workers and a Knowledge Graph, while Michael Hunger explored an IBM Attrition Dataset using Neo4j's Graph Data Science library. Additionally, Shaani Arya Srivastava published an article on how graphs may be a solution for environmental pollution, and the community has been discussing various topics related to graph databases, including reputation and success in art. The week also saw new videos recorded at the workshop becoming available on the Neo4j website, as well as tutorials on using Apache Hop to import data into Neo4j and Graphlytic's new export feature.

The author of the text has a background in data technology and law enforcement with Interpol, and was drawn to Neo4j after learning about its graph database capabilities and their potential to connect data. The author joined Neo4j's APAC team at a time of rapid growth and is proud to be part of the company's mission to help customers materialize their use cases and fulfill their data needs. With experience in amplifying customer voices, the author believes Neo4j addresses common challenges such as data silos, legacy platforms, and lack of skills, and has helped customers unlock insights and value through various use cases including NASA, Pandora Papers, financial services, and more. The author is excited to see where Neo4j's technology will take the world and invites others to "graph your own success story" by connecting data using tools like #IndexFreeAdjacency, visualization with #Bloom, and graph data science.

The new version of Neo4j GraphQL Java (1.3 and 1.4) introduces several exciting features, including integration with Spring-Boot DGS applications, allowing for the combination of schema augmentation and DGS's feature set. The library now supports type-safe APIs, better pagination, and alignment with neo4j/graphql.js. Additionally, default values for paging and sorting can be defined in the source schema, making it easier to query the API in a type-safe manner. The library also aligns its API with the official JavaScript version, with features such as using scalars for temporal types, separate sorting and paging into own input types, and improved testing capabilities.

Unleashing the power of knowledge is crucial for enterprises seeking a competitive edge, as turning data into knowledge remains an ongoing effort despite progress. Enterprises can benefit from connecting their data to create manageable and useful insights by starting small and gaining immediate value before expanding outward. Data management is an important use case for knowledge graphs, enabling companies like Lyft and Airbnb to thrive on data and identify customers in danger of churning or recommending personalized offers. Knowledge graphs form the foundation of modern data and analytics, capturing inherent relationships in data and unlocking predictive power through graph queries and algorithms. These insights can be applied to decisioning, where ultimate goals are to make better decisions whether human or algorithmic, and Neo4j offers a comprehensive solution for building actioning and decisioning knowledge graphs.

Neo4j AuraDB Free has been used in several HR applications and use-cases. The author of this post, who is the Head of Product Innovation & Developer Strategy at Neo4j, decided to explore an IBM Attrition Dataset on Kaggle as a way to experiment with AuraDB free. The dataset contains 32 columns of employee data, including attributes like job satisfaction, salary, and education level. The author imported the data into AuraDB free and created nodes for each employee, using the MERGE function to create new nodes if they didn't exist. They then extracted department and role information from the data, creating separate nodes for these concepts and connecting them to the employees. The author also explored temporal data in the dataset, converting date strings to zoned temporals and computing dates until which employees were employed. Finally, they discussed the potential for using graph data science libraries like GraphDB or NetworkX to compute similarity networks or node classification based on attributes, with the goal of identifying employees similar to leavers who have not yet left.

Hello everyone, it's week two of four for my run publishing TWIN4J. I'd like to take a moment to talk about Bonfire Night in the UK, which celebrates the foiling of a plot to assassinate King James I and his parliament on November 5, 1605. This tradition continues to this day with fireworks and bonfires being lit across the country. On to this week's events in Neo4j, we've had another jam-packed week of content that I'm delighted to deliver to your inbox. Featured community member Grant Beasley has been prolific in publishing blog posts on Neo4j, Rugby Analytics, and Prog Rock. He uses Neosemantics to import and explore 70's Prog Rock data, and his series on Graph Databases for Rugby Analytics explores the dataset using Python and Cypher. We've also welcomed Patrick Pichette to the Neo4j Board of Directors as an independent board member, with a background in finance and innovation. Additionally, we have articles on implementing Facebook social graphs using Spring and Neo4j, network visualizations with SigmaJS and GRANDStack, Neo4j Browser for functional visual validation, predicting mergers and acquisitions using graph-based deep learning, and more. Finally, my favorite tweet this week comes from Data Science Advocate Clair Sullivan, who publishes a new video every Friday on Bite-sized Neo4j for Data Scientists!

Graph data science is a science-driven approach to gain knowledge from the relationships and structures in data, typically to power predictions. Neo4j created the first enterprise graph framework for data scientists to improve predictions that drive better decisions and innovation by incorporating the predictive power of relationships and network structures in existing data. This approach helps answer previously intractable questions and increase prediction accuracy.

Graph: A Possible Solution for Environmental Pollution!` This article presents a graph-based solution model using Neo4j to address environmental pollution. The author, a Neo4j Certified Professional, aims to connect patterns of environmental pollution with industries responsible for it. The model uses nodes and relationships to represent cities, pollutants, disease cases, industries, raw materials, and manufacturing processes. A simple query helps reveal the industries linked to pollution and disease in a city. The author plans to expand the solution by incorporating additional factors such as industry domains, process links, and country-wise analysis. The article concludes with an invitation for suggestions and improvements from readers.

The article describes a new approach to visualizing graph data, called visual timeline analytics. It uses Neo4j, a graph database, and the KronoGraph toolkit to create an interactive timeline application that reveals the daily life of Twitter trolls. The application explores time-based graph data by transforming nodes and relationships into entities and events that unfold over time, providing insights into how connected events evolve. The toolkits used include KronoGraph and ReGraph for charting, which offer flexibility, scalability, and intuitive interfaces for end users.

The Neo4j Stack Overflow Sandbox is a new resource that allows users to analyze Stack Overflow questions, answers, tags, and users with the help of Neo4j APOC and Graph Data Science. This sandbox provides an interactive browser guide to get started with social network analysis without requiring a Neo4j environment setup or ETL process. The sandbox offers a 3-day trial period that can be extended up to 10 days. Users can explore data using Cypher query language, Load JSON, APOC, and Graph Data Science algorithms to create virtual graphs and compare tags. This resource is part of the Neo4j Developer Blog on Medium, with additional resources available for further learning.

The Neo4j team has made several exciting announcements and updates, including the launch of Neo4j AuraDB Free, a free version of their managed graph database service. This allows users to sign up and launch a fully-managed Neo4j database without any cost. Additionally, there have been updates on the Neo4j Connector for Apache Spark, which enables bi-directional streaming with Spark and treats Neo4j as any other streaming source. The team has also added a new Twitch dataset to Neo4j Sandbox, providing an opportunity to explore network analysis without having to download and install Neo4j. Furthermore, there have been updates on the GraphAcademy platform, which offers a series of courses designed to help users learn about Neo4j and its applications. The GraphAcademy platform includes courses on Neo4j Fundamentals, Cypher Fundamentals, Graph Data Modeling Fundamentals, and more. Additionally, the team has announced several new features and updates, including support for Open Source VR Server, Neo4j Integration in KNIME Analytics Platform, and a new blog post by Tomaz Bratanic visualizing thousands of flight routes between airports using sigma.js. The Neo4j team is also recognizing community members, such as Dr. Kim Martin, who has been exploring how to represent concepts relevant to her group as a Knowledge Graph using Neo4j.

Unleashing the power of knowledge is imperative for enterprises seeking a competitive edge, as most data landscapes are far from mature. Connecting data makes it inherently more valuable and provides dynamic context by enriching it with semantics. This enables reasoning about the underlying data, use of data confidently, and application in complex decision-making. Knowledge graphs can be created incrementally to solve practical challenges and serve additional stakeholders, providing agile structure that supports advanced AI and machine learning, ultimately leading to unlimited insights.

The text discusses Supervised Entity Resolution (ER) in Neo4j, a graph database. ER is the process of disambiguating data to determine if multiple records represent the same real-world entity. This is important for various industries such as online advertising, marketing, and law enforcement. The article explores how graphs can be used to represent associated information between subjects with paths made up of nodes and relationships. It also discusses the importance of feature engineering, data sampling, and hyper-parameter configuration in developing a supervised machine learning pipeline for entity linking. The authors demonstrate how to create a pipeline using Neo4j's Graph Data Science (GDS) library, which includes steps such as creating a graph projection, generating node embeddings, configuring the link prediction pipeline, training the model, making new entity linkage predictions, and querying resolved person information. The article concludes by highlighting the potential of ER in various industries and encouraging readers to experiment with the GDS library for their own projects.

At this year's NODES event, Neo4j's Chief Architect and CTO, Jim Webber, wrapped up the day with a keynote featuring interviews with customers and community members. One of the guests was Matt Cloyd, a civic technologist who discussed his work at the intersection of civic technology and graphs. He shared an exciting project on his horizon in conflict resolution and politically-related violence. Matt's journey into graph technology began when he realized that power mapping, a concept from his environmental sustainability course, could be applied to conflicts using graph databases. He developed a language called Aspen, which generates graph data and is specifically designed to be more humane than other graph technologies like Cypher. Aspen allows users to take notes about relationships in a conflict and have them automatically converted into Cypher for use in Neo4j. Matt's advice to newcomers to the graph community is to join the community, meet people, and be open to learning from others. He hopes that his work on graph-based analysis of conflictual events will help forecast and prevent violence in the future.

Neo4j is announcing Neo4j AuraDB Free, a cloud-based free version of their native graph database. The service allows users to create and manage databases without a credit card, ideal for learning graph databases or prototyping early development projects. Users can explore sample data, connect to the database from their program using drivers for popular programming languages, expand their graph horizons by importing custom datasets, and take their next steps in building applications. The service is cloud-native, requiring zero software setup, and provides an easy-to-use interface with a guided start-up experience, Cypher language tutorials, and community support.

The Neo4j Connector for Apache Spark now supports streaming operations, enabling bi-directional integration with popular streaming sources such as AWS Kinesis, Google Pubsub, and others. This feature allows customers to engineer data pipelines together with Graph Data Science, unlocking powerful use cases like Streaming ETL, data enrichment, trigger event detection, and machine learning. The connector reuses the Spark streaming API, creating a situation where Neo4j behaves just like any other streaming source, making it easy for developers to leverage their previous experience in batch operations. The feature is compatible with Neo4j AuraDB Free, Enterprise, Professional, as well as Neo4j Enterprise Edition and Community without limitations, supporting most modern versions of Apache Spark >= 2.4.5, including Spark 3.