July 2019 Summaries

Max De Marzi presents how decision trees are used to make near real-time decisions using a graph database, specifically with Neo4j. He uses the example of a nightclub entrance criteria where rules change at the owner's discretion. Max shows how to build a decision tree on top of a graph database using Java API traversal framework, which includes evaluators and expanders. The decision tree takes code as data and runs it in real-time, with up to 10 different rule nodes still delivering the same result performance. He also discusses building stored procedures, the decision path method, and the `isTrue` magical method that evaluates a rule's parameters and returns a Boolean answer. The use case is pharmaceutical drug recommendations where users build a decision tree based on research data, and Max shares his blog and open-source code repositories for further learning.

Graph technology plays a crucial role in enhancing the capabilities of artificial intelligence (AI) by providing context to AI applications. Narrow AI, which focuses on performing one task well, is becoming broader in its applicability as more general AI solutions are developed. Context is essential for making decisions, and AI needs to mimic human intelligence by incorporating context into its decision-making processes. Graph technology connects data and defines relationships, offering an effective means to empower the development of sophisticated AI applications. There are four main areas where graphs provide context: knowledge graphs, graph accelerated machine learning, connected feature extraction, and AI explainability. These areas enhance AI with related information, enabling it to make decisions closer to human intelligence. By providing contextual power, graph technology can give AI solutions a significant boost in their capabilities, making them more powerful over time.

This week the Neo4j community has been busy with various activities, including graph exploration from the Bloom canvas and a tutorial on building a Graph App in vanilla JavaScript. Max De Marzi also released a procedure that computes parallel k-Hop counts using Neo4j's Internal API. Additionally, Nicole Forrester is featured as a community member for her design work on projects such as the Global GraphHack 2019 t-shirts. The community has also seen releases of a Scala Cypher DSL and a library that packages functionality for working with Neo4j and R. Furthermore, Will Lyon and Amy Hodler published their slides from their Graph Algorithms talk at OSCON 2019.

David Bader, a researcher at Georgia Tech, discusses the university's use of Neo4j graph database technology to solve real-world problems such as improving transportation, security, and insider threats. He highlights Neo4j's capabilities in rapidly asking questions and visualizing results, making it an ideal platform for day-to-day work with graphs. Bader also shares his vision for the future of graphs, which includes preventative analytics and incorporating massive streams of datasets to make decisions in real-time. The researcher expresses his enthusiasm for attending GraphConnect conferences, where he sees a growing community of graph enthusiasts from various industries, including finance, retail, security, and government contractors.

Neo4j is hosting a hackathon, Neo4j Online Developer Expo and Summit (NODES 2019), where community members of all experience levels can participate. Hackers will have four weeks to work on their project submissions, with the projects being published one week before the live broadcast and allowing global voting by the Neo4j community and staff. The theme will be announced in September, and top three winning teams will receive various prizes including travel, registration, private-access to exclusive events, meet-and-greets with Neo4j executives, poster sessions, and blog post features. Registration for hackers is required, with a deadline of August 30th, followed by a four-week hacking period, community voting from October 1st, and winners being announced on October 10th during the NODES 2019 keynote.

AIG was seeking to consolidate their processing, understand how teams contributed to goals, and make changes efficiently while seeing the impact upstream and downstream. They modeled their Integrated Object Environment (IOE) after Zillow's approach, focusing on holding information that flows both upwards and downwards. AIG collaborated with Tom Sawyer Software to create a tool efficient, easy to update, and user-friendly. Neo4j was chosen as the best-in-class graph database due to its powerful features. The IOE consists of four components: attributes, diagramming, standards, and a data visualization layer. Attributes are used to run queries, analyze dimensions, and optimize processes. Diagramming is essential for understanding complex systems, while standards ensure consistency. The system enables users to visualize topology, identify risks and contributions, and make informed decisions. AIG's focus on staying focused on what the tool can do that saves money is crucial, rather than getting bogged down in its potential capabilities. The IOE measures propagation of risk but also propagation of contribution, using attributes to analyze different dimensions and optimize processes. It becomes an entry point into artificial intelligence.

This webinar provides an overview of different Neo4j versions, discussing their features and services, and presenting alternative licensing options such as open source and commercial licensing. The free open source version is suitable for individuals, startups, and academics, while paid commercial options are available for larger organizations. The webinar aims to help attendees choose the most suitable Neo4j version for their needs.

The text discusses the importance of trustworthiness in AI standards efforts, particularly in the context of reliability, fairness, and explainability. Graph technology is highlighted as a key enabler for adding contextual information to AI systems, enabling them to be more reliable, fair, and trustworthy. This approach allows for data lineage tracking, guarding against input data manipulation, and avoiding excessive reliance on correlated data. Contextual information also reveals potential biases in existing data and enables situational/contextual fairness. The text concludes that graph technologies are essential for promoting AI aligned to key American values of accountability, fairness, and trust, and that they offer a more human-friendly way to evaluate connected data.

This week, the Neo4j community has been actively engaged with several exciting developments. Anurag Tandon shared his expertise on Near Natural Language Search in Bloom, a graph visualization tool, and introduced its search capabilities, including basic search terms, search phrases, and pattern-based search. Additionally, David Allen announced the release of Neo4j Enterprise Causal Clustering on Pivotal Container Service (PKS), making it easier to spin up a clustered, highly available graph database on top of Kubernetes. The community has also been exploring the use of Siren, an investigative intelligence platform, with Neo4j, and Dave Fauth is starting a series of posts connecting Neo4j, MySQL, and Kafka. Furthermore, Dr. Vlasta Kus, a data scientist at GraphAware, was featured as this week's community member, sharing her expertise on graph-based machine learning and presenting talks on AI and NLP at GraphTour Chicago. The community is also working on new releases of the Graph Algorithms Library, introducing procedures to compute memory requirements, fine-grained control over concurrency settings, and usability and bug fixes.

This week, a video was published by David Allen on deploying Neo4j on Google Kubernetes Engine (GKE) via GCP Marketplace. The video explains how using Neo4j on GKE allows developers to focus more on their business applications and less on cluster provisioning and maintenance tasks. The video is part of the Neo4j YouTube channel, which offers a weekly stream of graph technology content. To access all available videos, viewers can subscribe to the channel.

For those interested in learning more about graphs within the financial services industry, this session reviews typical data types, illustrates graph formations, and discusses use cases such as anti-money laundering, fraud detection, and regulatory compliance. The presentation highlights various companies already utilizing Neo4j's database for different purposes like network management, search, master data management, price comparison, and identity access management.

Graph technologies have increasingly been integrated with machine learning and artificial intelligence solutions, enabling better predictions, flexibility, and explainability in various applications. By incorporating context into AI models, relevant information can be added to existing data, resulting in improved predictive accuracy. This is particularly evident in the fraud detection and prevention market, where graph algorithms are used to reveal predictive patterns, find unusual behavior, and score influential entities. Additionally, contextual awareness and situational flexibility are crucial for AI systems, enabling them to adapt to new situations, reduce failures, and improve resilience. The use of graph technologies has a broad applicability across various industries, including financial services, software firms, logistics companies, retailers, airlines, telecommunications firms, and hospitality companies. By incorporating context into AI models, robustness and reliability can be enhanced, leading to more accurate predictions, flexibility, and improved decision-making.

This week, Greta Workman interviewed Mark Grover, Product Manager at Lyft, about Data Discovery at Lyft and the Amundsen project, an open source data discovery application that uses graphs and Neo4j to improve productivity for practitioners interacting with data. Registration is now open for GraphConnect 2020, a conference featuring talks on Neo4j usage from around the world. Additionally, Kristel Nieuwenhuys and Ronald Willems discussed their tool Software Parade, which helps track applications in a micro services architecture, while Max De Marzi shared how to solve link prediction problems using motifs in Cypher. The Neo4j community also welcomed Chris Skardon's beta launch of a Neo4j integration with Microsoft's SQL Server Integration Services (SSIS), and Claus Engel-Christensen introduced the Neo4j Client for Clojure.

Convergys, a large customer service outsourcing company with over 120,000 employees worldwide, was struggling to meet the EU's General Data Protection Regulation (GDPR) deadline. They were looking for a solution to handle their vast amounts of structured and unstructured data across multiple client systems, which posed a significant challenge. The company discovered graph technology, specifically Neo4j, as a potential solution to this problem. After meeting with several people who had built solutions on top of Neo4j at GraphConnect 2017, they selected Neo4j along with another company, CluedIn, to develop a custom solution. They found that using Neo4j enabled faster data ingestion and processing than expected, which was surprising given the complexity of their data. The company has since seen operational results improve with the use of graph technology, enabling better decision-making through connected data. In the future, Convergys believes that graph technology will be widely adopted across various industries to achieve better operational results.

GraphConnect 2020 is a major gathering of graph database developers, architects and CTOs taking place in New York City on April 20-22. The event will feature presentations by speakers from around the world showcasing various use cases, industries and innovations where Neo4j has had a significant impact on their organizations. The conference offers something for everyone, including training workshops, one-on-one clinics, a hackathon, exhibits, and more. Early bird pricing is available until a certain point, making it an ideal time to register before the sale ends at its lowest price. Attendees can expect a relationship-rich community where they can connect with others in the field.

The presentation discusses how a vascular surgeon, Christopher Wixon, transitioned from feeling burnt out and frustrated with electronic data entry tools in the healthcare industry to creating a knowledge graph that integrates graph technology with health information technology. The graph technology aids both physicians and patients by increasing expediency in retrieving information and diagnosing patients. The speaker believes that graphs could be extended to represent the entire human condition, but acknowledges the challenge of building such a complex system. The presentation showcases examples of how the knowledge graph can aid orthopedic surgeons and other medical professionals in documenting patient visits and making diagnoses. It concludes by highlighting the potential benefits of the knowledge graph, including improved outcomes, reduced costs, and enabling less experienced clinicians to function at the top of their credentials.

The complexity of modern consumer products, such as cars, is staggering, with numerous communication buses, control units, software components, and mechanical parts working together to create a product that meets regulatory and safety requirements. To manage this complexity, manufacturers rely on data silos, but these can lead to partial views of the product's information. The next step in managing data effectively is through AI and knowledge graphs, which create knowledge context from relevant information, enabling companies to make better decisions and automate processes. Knowledge graphs are dynamic, linked heterogeneous data, and intelligent metadata enable data mining and simplify navigation, allowing for deeper searches and answering new questions. By mapping the product's DNA using a knowledge graph, manufacturers can design a map of functions and trace dependencies, identifying clusters and loops in the system design, optimizing project planning, and maintaining overall functionality. Ultimately, AI and knowledge graphs provide increasingly dynamic, relational, and connected representations of data, allowing for improved expediency in deciphering the product's DNA.

Graph technology is a natural fit for AI and machine learning projects across any industry, as it represents connected data and analyzes relationships in a way that treats the connections between data equally important as the data itself. Graph technologies store and use data by showing how each individual entity connects with or is related to others, acting as a fabric for our data imbuing it with context and connections. This approach enriches data making it more useful and meaningful, similar to how stars become more significant when part of a constellation. Graph technologies were originally custom-built for companies like Facebook and Google but have since been developed into graph databases that allow organizations to quickly traverse millions of data connections per second. The technology is now used across multiple industries including government, financial services, healthcare, retail, manufacturing, and more, with applications such as fraud detection, cybersecurity, real-time recommendations, network & IT operations, master data management, and customer 360. Graph technologies have also been recognized as a major leap forward in machine learning, enabling better accuracy, flexibility, and explainability in predictive models and decision-making processes.

This week, Neo4j has launched a series of videos on Neo4j Bloom, the graph visualization tool, with Anurag Tandon leading the charge. Adam Cowley has also shared his expertise on text analytics using Neo4j and Cypher. The company has also introduced new project pages for its recently launched Neo4j Labs initiative, which provides a platform for innovation and quick iteration. Meanwhile, Jennifer Reif, a key member of the Developer Relations team at Neo4j, is being featured as a community leader, having traveled extensively to teach users about graphs and contributing significantly to the development of Neo4j resources. Additionally, new content has been released on topics such as calculating TF-IDF scores in Neo4j with Cypher, learning about research connectivity using Neo4j and W3C annotations, choosing Kubernetes for cloud infrastructure management, and more.

Our Managing Editor showcased a video about Swedish Midsummer last week, which is close to our company's Swedish roots. The Neo4j Developer Relations team produced an Intro to Graph Databases video series that helps you become a graphista in no time, covering the evolution of databases and topics such as use cases, data migration and Cypher. You can catch all the videos on the Neo4j YouTube channel, which is updated weekly with new content, or subscribe to our channel for more graph tech goods.

The video discusses Elisa's experience building a live inventory and topology store using Neo4j, a graph technology platform. This allows for accelerated deployment of network automation solutions, modernizing the network. Graph technology is also used in other areas to build adaptive solutions for major telco providers, future-proofing their operations. The webinar provides insights into how these technologies can be applied to improve network management and deployment.

Hume is a platform that aims to transform data into real wisdom and inspires quantifiable change by leveraging knowledge graphs, entity recognition, concept enrichment, and topic modeling. The platform uses Neo4j as its underlying graph database and provides tools for collecting, extracting, and enriching data, including named entity recognition, entity and concept enrichment, and topic modeling. Hume's goal is to enable machines to collaborate with humans in understanding and utilizing large amounts of information efficiently and effectively. By converting data into knowledge, Hume enables businesses to deliver better products, analyze production chains, and improve services for their end-users. The platform has the potential to revolutionize how information is shared and utilized, making it more accessible and useful for people.

The National Institute of Standards and Technology (NIST) aims to create a plan for Federal engagement that supports reliable, robust, and trustworthy artificial intelligence technologies by establishing a common technical standard. However, AI must be guided not only by technical standards but also by ethical standards, as context is crucial in understanding the peripheral information relevant to specific AI applications. Graph technology is proposed as a state-of-the-art method for adding and leveraging context from data, providing a powerful foundation for AI. By incorporating context, AI can make human-like decisions that are more situationally appropriate, ensure explainability and transparency of decisions, and prevent biased outcomes. The lack of context in AI standards results in subpar outcomes, uninterpretable predictions, and less accountability, as seen in cases like Microsoft's Twitter bot Tay learning offensive language and Amazon's AI-powered recruiting tool showing bias against women candidates. Understanding the larger frame of reference is critical for identifying whether an AI project has gone off-course, and explainability is essential for accountability, particularly in nuanced situations.