September 2018 Summaries

This week in Neo4j has seen the introduction of Deep Feature Learning for Graphs with the DeepGL algorithm, which is designed to keep memory usage low and returns feature names for model understanding. There have also been presentations on Customer Journey Analytics using Neo4j, Decision Trees in Neo4j, Data Vault implementation, and a GA release for Neo4j-OGM 3.1.3. The community has also celebrated the work of featured member Reshama Shaikh, who is a bio statistician and data scientist, organizer of Women in Machine Learning and Data Science NYC meetup groups, and wrote a blog post about her GraphConnect experience. Additionally, there have been announcements for Neo4j 3.4.7 enterprise causal cluster update on AWS Marketplace, the launch of JQAssistant Dashboard, and upcoming events.

The text discusses how Blackstone Technology uses Neo4j to visualize and analyze its government client's enterprise architecture. The company connects various IT assets using information from diverse systems, allowing decision-makers to see opportunities for optimization and cost reduction. Patrick Elder and Jessica Dembe, Front-End Engineers, use Neo4j to stitch together disparate data sources and create a visualization of the organization's connections. They found native visualization capabilities in Neo4j attractive and exciting, as it allowed them to quickly put together a visual representation of their client's enterprise architecture. The team has seen impressive results, including a "Wow" factor from their clients, who were impressed by the speed and efficiency of putting data into Neo4j. Looking back, they would start with iconography for different investment areas and consider starting with analysis instead of visualization. They believe graph technology will be valuable in the federal space, particularly for document analysis and operationalizing Natural Language Processing capabilities. Overall, GraphConnect has provided a great experience for Blackstone Technology, allowing them to learn from more mature projects and bigger datasets.

The ability to visualize connections on maps has been highly valuable for centuries, from Chinese astronomers plotting 1,300 stars over 1,000 years ago to modern-day use cases such as monitoring traffic flow or identifying routes used by persons of interest. Neo4j's graph database and KeyLines' JavaScript SDK can help exploit geospatial information, providing a wealth of technology to gain insight from links between locations. Dashboards offer the "big picture" at a glance, allowing analysts to drill into specific locations for detailed analysis. Maps can also be used to give new perspective on old information, such as in fraud investigation or cyber threat intelligence, by visualizing connections and relationships between data points. KeyLines' flexible approach allows users to choose from various map tile providers, projection systems, and plugins to integrate with the software, making it easy to visualize graph data on a map that suits their needs.

GraphConnect 2018 was a successful event that brought together Neo4j customers, partners, celebrities, prospects, skeptics, employees, and community members for conference sessions, training workshops, and a massive hackathon. The event featured live-streamed keynotes by Neo4j CEO Emil Eifrem and Hilary Mason, the GM of Machine Learning at Cloudera, which were available to watch now. Neo4j 3.5 was announced and is now available for preview, with new features aimed at supporting AI and machine learning applications. The Graphs4Good program was launched to connect and enable those working with data to make a positive impact. The event also recognized innovative graph technology applications through the Graphie Awards, celebrating success in connected data across multiple categories. Video recordings of the event will be uploaded to the Neo4j YouTube channel, and it's never too early to think about next year's GraphConnect 2019.

Risk modeling in banking is a complex task that requires tracing data connections across various investment baskets, holdings, financial instruments, and pricing data. The Financial Reform Transparency Bill (FRTB) requires banks to decompose risk models into their individual components and trace back through time to available pricing and position information. This process involves identifying relevant data, understanding data sources, and calculating risk factors that affect all upstream information dependencies. Effective internal risk models require a strong foundation in data governance, which is essential for risk aggregation, reserve calculations, and required reporting. Modern graph technology, such as Neo4j, provides a robust framework for building and testing internal risk models that can trace many layers of dependencies and adapt to changing market conditions.

This week in Neo4j saw the opening keynote of GraphConnect 2018, where Hilary Mason and Emil Eifrem presented on Machine Learning and Artificial Intelligence, highlighting its applications in solving new problems. The APOC library was also released with support for custom procedures, web scraping, and refreshed documentation. Additionally, Graphs4Good was announced as a program to showcase graph-powered projects that promote positive social change, while Let's Encrypt was used to set up valid SSL for protecting data in transit. New releases of the Graph Algorithms library included similarity functions, balanced triads, and Louvain Phase 2, with support for calculating similarity between nodes and intermediate cluster assignments. The Talking Kotlin podcast featured an interview with Michael Hunger discussing graph databases and his work integrating Neo4j with GraphQL and Kotlin. Next week's events include a talk on September 27th, and the tweet of the week highlights inspiration from Hilary Mason's talk on AI and ML at GraphConnect 2018.

The Neo4j Graphie Awards celebrate innovative graph technology applications across multiple categories. The awards recognize successes in connected data, not only for Neo4j customers but also across the entire Neo4j community and ecosystem. This year's winners include Adobe Behance, Microsoft, Comcast, eBay, Pitney Bowes, the German Center for Diabetes Research, DXC Technology, Convergys, Graphen, Juit, and Iryna Feuerstein, who was recognized as a Community MVP. The awards aim to honor innovative graph use cases and encourage more nominations in future years, with the award categories potentially changing.

In 2000, Emil Eifrem co-founded Neo4j after conceiving the first property graph model on a napkin during a flight, leading to the creation of the world's first graph database. As detailed in his blog post, the impact of Neo4j has extended beyond business, with the graph technology being used to address global challenges such as climate change, cancer research, and financial crime. The newly launched Graphs4Good program aims to support projects that leverage graph technology for social good, with past successes including investigative journalism projects like the Panama Papers and data-driven research in fields like medicine and space exploration. Neo4j's community-driven approach has fostered collaborations with organizations like NASA and the German Center for Diabetes Research, while initiatives like data science camps for women highlight the company's commitment to expanding the role of technology in positive societal change.

The text discusses transforming unstructured data into actionable knowledge using a graph database and machine learning techniques. It highlights the importance of representation, knowledge learning and construction, insight and wisdom gained from connected data, and the use of various algorithms such as Word2Vec, named entity recognition, probabilistic topic modeling, and sentiment analysis to extract value from large datasets. The text also describes how these techniques can be integrated into a platform called Hume, which uses a graph database and machine learning to transform data into searchable, understandable and actionable knowledge.

The text discusses how data visualization can improve the understanding of query results and recommendations from Neo4j, a graph database. It highlights the importance of rich visualizations that include more information about elements on the screen, such as numeric data, colors, gauges, and badges, to provide an enhanced user experience. The text also emphasizes the need for sophisticated visualization solutions with level of detail rendering, versatile layout algorithms, and interaction capabilities to make complex diagrams easier to navigate and understand. Additionally, it introduces yFiles, a graph visualization library that can be used in conjunction with Neo4j to create interactive and informative visualizations.

The GraphHack event will be held at the Stack Overflow office in New York City on September 22nd, featuring Neo4j's integrations with other popular technologies. The event is open to anyone from the Neo4j community, regardless of experience level, and includes workshops for those new to Neo4j. Teams will build applications using Neo4j and other listed technologies in a "Buzzword Bingo" format, with prizes including Oculus Rifts, GoPros, Bose SoundLink Color Bluetooth Speakers, and more. The event schedule includes optional workshops, hacking time, presentations, and a GraphHack cocktail hour. Participants must join teams, present at the end of the day for a chance to win, and provide information on their used technologies. Guests will need a photo ID to enter the building.

This week in Neo4j has seen various updates and releases, including new AMIs for AWS and a Java Driver release. The company has also announced the release of its Bloom data visualization tool, which allows users to visualize graph data in a natural language search interface. Additionally, there have been several community member spotlights, including Scott Sosna, who is using Neo4j to explore different open datasets. The week has also seen new articles and blog posts on topics such as personalized page rank, solving the bucket-filling problem, and deep text understanding. Furthermore, the Graphistania podcast featured an interview with Karin Wolok, Neo4j's Program Manager of Community Development and Enablement. Upcoming events include GraphConnect NYC 2018 and a poll on the Neo4j forum asking about where users run their Neo4j instances.

The estimated volume of money laundering worldwide is approximately $2 trillion, with Germany being a significant contributor at $100 billion. Governments and companies are implementing stricter regulatory requirements to combat this issue, affecting corporations of all sizes. Effective Anti-Money Laundering (AML) compliance requires more than just technical measures, but also risk analysis, transaction monitoring, employee training, and documentation. A comprehensive solution is crucial for managing complex data networks, which can only be handled with an IT system like KERBEROS, built on native graph database Neo4j. This solution provides flexibility, allowing developers to adapt to evolving legislation and regulations, and enables business users to define and run database queries without coding expertise. The KERBEROS solution is a prime example of how graph technology can overcome rigorous compliance challenges such as multi-layered risk management regulations in AML.

Graph visualization is a powerful tool for working with big data, but its intuitive nature can be limited by screen size constraints. The author's startup leveraged graph signatures to analyze social network accounts, but struggled with 2D layouts that failed to reveal the full structure of the data. Switching to 3D graph visualization enabled better separation of clusters and improved visual understanding of complex patterns. The use of virtual reality (VR) for data visualization has shown promise, allowing users to gain situational awareness of complex patterns without constant zooming or jumping between views. While VR is still in its early stages, it holds great potential for enhancing the visualization experience, particularly with advancements in augmented reality (AR) headsets that will soon overcome current limitations. The future of data visualization is exciting and full of opportunities for innovation.

To provide users with a more useful experience from big data, it's essential to apply techniques at different stages of the data funnel. Filtering in Neo4j reduces noise by removing as much data as possible early on, utilizing Cypher queries and KeyLines' integration for effective visual filtering. Aggregating data cleanses duplicates and errors, while also considering data modeling to remove unnecessary clutter. By creating a clever visual model with a small proportion of the original data, users can simplify complex patterns and gain insights. Finally, using filters, combining, and pruning techniques declutters the chart, and automated graph layouts help uncover patterns and anomalies, making it easier for users to find answers in their data.

The Fundamental Review of the Trading Book (FRTB) regulations aim to create specific capital-reserve requirements for bank trading desks based on investment-risk models, with the goal of maintaining solvency through market downturns and avoiding governmental bailouts. FRTB mandates have led banks to build a firm foundation for future risk management and compliance applications, optimizing reserve ratios and maximizing available capital while driving investment profits. The regulations raise Basel reserve requirements, focusing intensely on the trading desk and requiring banks to develop or approve internal risk models to calculate capital-reserve requirements. Internal model approval is crucial, as it demands a bank's ability to trace data dependencies through complexity, which can be effectively captured using a graph database like Neo4j.

This week in Neo4j covers various topics including modeling complex financial instruments using the graph database, loading streaming data into Neo4j, solving a Rubik's cube and shuffling a pack of cards using random walks, and using Kettle as part of an ETL process. The featured community member is Ron van Weverwijk, a long-time member of the Neo4j community who has given training sessions and contributed to the APOC library. The Graphistania podcast also features an interview with Johannes Unterstein, a Software Engineer working on the Neo4j Cloud product, discussing his experiences and insights on managed services and cloud computing. Additionally, there are updates on various projects and tools such as Graphileon's new bookmarking functionality, Joy Chao's blog post on triadic closures, and the Neo4j Spark Connector for loading JSON data into Apache Spark DataFrames.

Hilary Mason, GM of Machine Learning at Cloudera and Founder & CEO of Fast Forward Labs, will deliver a morning keynote speech titled "The Present and Future of Artificial Intelligence and Machine Learning" at GraphConnect 2018. She will share her real-world experience with AI and discuss the benefits and future potential of AI applications in organizations. In contrast, Stephen O'Grady, Principal Analyst and Founder of RedMonk, will deliver a closing keynote speech titled "What Will You Build, and Why? The motivations, ethics, and career opportunities of modern application development". His session will focus on personal and moral questions developers need to answer when choosing jobs and projects.

The Cypher Philly initiative is an open source project that aims to empower citizens, journalists, data scientists, coders, and creatives with the ability to harness open public data for civic good. The team was inspired by the impact of data on investigative journalism and wanted to simplify the process of telling data-driven stories using open public data. They have gained local sponsorship from Linode and Azavea, and are collaborating with various organizations and communities to build and expand their reach. The project provides a collection of digital tools and methods for finding, scraping, importing, and storing data, which are freely available on GitHub. Participants can contribute to the projects by completing tasks and issues assigned to them, and collaborate in real-time meetings to address major civic issues such as gerrymandering and representation.

Graph theory, a type of math that doesn't use numbers, is used to analyze and understand the structure of graphs, which are collections of nodes and edges. Triadic closures refer to the observation that if two nodes are connected via a path with a mutual third node, there's an increased likelihood of the two nodes becoming directly connected in the future. This concept is useful for predictive modeling in social networks, fraud detection, and other applications where understanding relationships between entities is crucial. Structural balance is another aspect to consider in the formation of stable triadic closures, which involves analyzing the quality of relationships involved in the graph. Local bridges are a tie between two nodes that are not otherwise connected or share common neighbors, and predicting these weak links can be useful for tasks like job search recommendations and data lineage tracking. Understanding graph theory and its applications can help achieve business goals, and there are many resources available to learn more about it.

In this article, we explore how real-time recommendations support various use cases, including product recommendations and logistics. A graph database is used to store customer purchase history, which can be queried to suggest popular products or personalized recommendations based on individual customer behavior and social connections. The query examples demonstrate how Neo4j's Cypher language can be used to extract insights from the graph data, such as recommending historically popular purchases made by a customer and suggesting products purchased by their friends and friends-of-friends. By leveraging graph technology, organizations can incorporate customer feedback, adjust for seasonal trends, and provide personalized recommendations in real-time without complex coding or relational JOIN issues.

This week in Neo4j has been exciting with various community-driven projects and announcements. Maxim Salnikov, a Full-Stack Engineer at ForgeRock, was featured as the community member of the week for his contributions to building relationship-based digital identity platforms using Neo4j. The community has also seen the launch of a new Neo4j forum with trending activity feeds built using GraphQL and Neo4j. Additionally, several community members have shared their experiences and expertise on various topics such as NetSCAN community detection algorithm, Neo4j to Contentful in Elixir, and tips for passing the Neo4j Certification exam. The community is also building a dating site with messaging functionality, which has been documented in a series of blog posts. Looking ahead, there's an upcoming event on September 5th, 2018, where Chris Eyre will be giving a talk and workshop about the Neo4j to Contentful library in Elixir.