July 2018 Summaries

The text discusses how graph technology has made data modeling more accessible to a wider audience, allowing anyone with basic knowledge of graphs to create rudimentary data models. However, this increased accessibility also means that data modeling design can go wrong, and weak data models can lead to poor application performance. The author will guide readers through the basics of graph technology and provide examples to illustrate common mistakes in data modeling. A key example is a fraud detection application analyzing users' email communications, which highlights the importance of capturing relevant elements and activities in a graph data model. The author provides two iterations of the data model, each addressing specific issues, such as adding nodes for emails and relationships to track sender and recipient information. The final iteration includes tracking replies and forwards, demonstrating how a robust data model can provide valuable insights into complex problems.

Forward-looking banks are uniting data silos into an information foundation for building innovative applications that provide extreme visibility and deep analytical insights, improving compliance efforts and day-to-day decision making. The Basel Committee's BCBS 239 regulations aim to improve risk reporting by establishing standards for recording and tracing financial transactions, aggregating risk data, and adjusting capital ratio requirements accordingly. To comply with these regulations, banks must utilize data governance and integrated data taxonomies, generate accurate and consistent risk data, manage and report risk in a precise and auditable manner, and address key data challenges such as data lineage, silos, terminology differences, legal entity identifiers, and data consistency and latency.

This week in Neo4j has seen the release of a new version of the Graph Algorithms library with support for the Random Walk and Personalized PageRank algorithms, as well as an eBook on graph algorithms that covers both theoretical and practical aspects. The community has also been active with various projects and contributions, including Global Witness publishing their final report on graphing the UK company ownership register with Neo4j, David Mack sharing his research on machine learning and graph-based architectures, and several open-source stack for software analysis and visualization being developed. Additionally, there have been some interesting projects to play with, such as WikiLink, Hospital-Organ-Transplant-API, and Family History App, and a Graph Algorithms eBook has been released that covers both theoretical and practical aspects of the topic.

IntelliTag's Symmetry, built using Neo4j, is a total property management system for casinos that integrates disparate data sources to create a holistic view of guest movements and optimize casino operations. The product uses Neo4j's graph database to normalize data from various sources, enabling real-time analytics and insights. IntelliTag chose Neo4j due to its ability to quickly integrate and analyze large amounts of data, particularly in the financial services sector where speed is crucial for decision-making. The company has seen surprising results with Symmetry, including aha moments for users and executives who can now see a unified view of their properties and make informed decisions. As the gaming industry continues to evolve, IntelliTag believes that Neo4j's graph technology will play a key role in enabling machine learning, artificial intelligence, and governance, particularly in regulated markets. The partnership between IntelliTag and Neo4j has been successful, with Neo4j providing support and helping the company penetrate verticals.

I've summarized the text for you. Here's a neutral and interesting paragraph covering key points: A female engineer recounts her experience of being underrepresented in the field, where only around 5% of computer science students are women, despite progress being slower than anticipated. She attributes this to lack of female role models, colleagues, and balancing work-life as a parent, not specifically due to being a woman. However, she finds joy and confidence among like-minded women at Pink Programming events, which aims to create an inspiring environment for all women in tech. Her company partners with Pink Programming, hosting events that promote female engineers and data science education, fostering connections and friendships among participants. The partnership has expanded their networks and encouraged women to pursue IT education and careers.

Telia Zone is a router used in approximately one million homes in Sweden, connecting to the Internet and hosting causal clusters with Neo4j to graph actions taking place in and through the routers. The company has expanded its use cases beyond basic connection, using the graph to determine new capabilities for the router, such as receiving messages when children arrive home or automatically generating playlists for Sonos speakers. Telia uses Kubernetes to scale Neo4j, hosting causal clusters with Google Cloud Platform and running Node.js apps. The platform is expanding into externalizing intelligence gathered from consumer data as a B2B offering. With its graph database, the Telia Zone allows users to ask questions like "What other devices are running this app?" or "What kind of relation did this device have to this other zone?", enabling business intelligence and unique insights such as predicting Christmas present trends. The underlying technology is found in the APIs on premiumzone.com, positioning it as a technology that will be expanded outside of Telia's footprint soon.

This article discusses the basics of data modeling, particularly in the context of graph databases, and highlights the differences between relational and graph data modeling. The author shares their personal experience with learning data modeling using a relational database, which they found to be a difficult and frustrating process. They argue that this is not inherent to data modeling itself, but rather a result of using an RDBMS as the default choice. The article introduces the concept of graph data modeling, which the author finds to be more intuitive and easier to work with. It explains how the graph data model can adapt to changing business and user needs, making it a better fit for rapidly evolving applications. The author concludes that relational databases are suitable for well-understood, minimally changing data models, but graph databases offer a more agile and flexible solution for new or uncertain projects.

Neo4j is a highly scalable, native graph platform that empowers businesses to rapidly build next-generation service assurance solutions. It delivers real-time insights into data relationships and enables fast writes of dynamic topology and lightning speed traversals. With its schema-less model and built-in high availability features, Neo4j allows organizations to continually improve their network solutions without requiring a rewrite of their data model. Companies like Zenoss, Cisco, Orange, and Telenor are leveraging Neo4j to differentiate their solutions, go to market faster, and provide performance at scale. By using Neo4j, businesses can prototype faster, complete rapid proof of concept projects, execute quickly in production, and iterate and expand their solutions with ease.

The author, Head of Product Innovation & Developer Strategy at Neo4j, explores the possibilities of visualizing graphs in 3D using WebGL and the 3d-force-graph library. They start with a simple example using the Game of Thrones interaction graph and demonstrate how to load data from a Neo4j database using the JavaScript driver. The author then shows how to add color, captions, and weights to nodes and relationships, as well as implement incremental loading, particle effects, and cluster coloring. The visualization is made possible by the 3d-force-graph library, which provides a Graph API and various options for customizing the graph's appearance.

This week in Neo4j has seen significant progress and announcements from the community. David Fox, a member of Adobe's backend infrastructure team, shared his experience of moving their activity feed feature from Cassandra to Neo4j, highlighting the reduction in dataset size and exponential decrease in developer-operations staff hours required. Additionally, Neo4j has launched a commercial Kubernetes application on GCP Marketplace, allowing users to easily deploy Neo4j's native graph database capabilities for Kubernetes directly into their GKE-hosted Kubernetes cluster. Emil Eifrem, Neo4j's CEO, was interviewed on The New Stack Makers Podcast, discussing the history of Neo4j and its vision for Machine Learning and graphs. The first alpha release of the Go driver has also been made available, providing a familiar API for users of other language drivers. Furthermore, community members have shared their experiences with APOC procedures, including creating nodes and relationships dynamically with dynamic data, and analyzing graph dependencies using centrality algorithms from the Neo4j Graph Algorithms library. Finally, upcoming events include Mark Needham's talk on Neo4j Quick Graphs: Extracting Taxonomies, Strava, Wikipedia, Python Dependencies, scheduled for July 25th 2018.

The University of Washington's IT team built its own metadata tool using Neo4j to connect all their metadata and handle the ever-changing schema of the university's data. They found that Neo4j gave them the ability to connect any node to any other node and show that visually, providing context for their metadata. The team uses Neo4j as a metadata repository to stitch together information for the enterprise data warehouse and BI tools, making it easier for end users to visualize and get context for metadata. They have mixed the metadata with other data sources, such as security information and organizational structure, to create new insights and reports. Looking back, they would change their data model to avoid versioning complications that made Cypher queries more difficult. The team believes that better UI tools are needed to allow end users to analyze metadata in a more visual and user-friendly way.

Capgemini argues that using multiple databases improves the ability to unlock business benefits from data by allowing for a more flexible and tailored approach to data analysis. The company suggests starting with one database that partially meets all analysis needs, then adding other databases such as graph or NoSQL to address specific bottlenecks. This approach can help achieve faster query performance, improved productivity, increased insights potential, and better governance, while also mitigating the costs of increasing IT spend, integration complexity, and requiring diverse skill sets. However, it's recommended to work iteratively, starting small, and adding databases as needed to test performance needs and justify the use of additional technologies.

The Neo4j Graph Platform is now available within a commercial Kubernetes application on the Google Cloud Platform Marketplace, providing customers with easy access to its native graph database capabilities. This new offering allows users to deploy Neo4j alongside other workloads in their GKE-hosted Kubernetes cluster, using a "Bring Your Own License" model for enterprise edition licenses. The introduction of this application expands the possibilities for Kubernetes users to combine graph technology with existing applications, such as those generating recommendations or building 360-degree customer views. The Google Cloud Platform Marketplace offers flexibility and innovation through its multi-cloud and hybrid-first philosophy, allowing customers to easily adopt new technologies and oversee the full lifecycle of a solution. With this announcement, Neo4j supports containerization technology, including Docker, and enables users to pair it with existing applications or install other Kubernetes marketplace applications alongside Neo4j.

In business and life, relationships matter more than individual skills or competencies. However, when it comes to data, we often overlook these connections, focusing on individual data points rather than their relationships. Relational databases are great for handling discrete data points but struggle with complex queries that require connecting multiple pieces of information, leading to poor performance. NoSQL databases can store connected data but have limitations in handling reciprocal queries and may not be efficient enough for deep and complex queries. In contrast, graph databases store data relationships as relationships, allowing for flexibility and efficiency when it comes to querying and adding new nodes and relationships without compromising the existing network or expensively migrating data. Graph technology is particularly effective at handling connected data, making it a crucial tool for mission-critical insights and nimble business agility.

The network is a complex, interconnected ecosystem that can be effectively modeled using graph database technology. This approach provides a unified view of the infrastructure and topology, breaking down silos and enabling real-time decisions to be made about network services. A native graph approach, such as that provided by Neo4j, naturally captures relationships between data and models network complexity, outperforming relational databases in querying and processing such complexity at scale. This enables service assurance, allowing for performance and predictability, and rapidly diagnosing failures by correlating and tracing user complaints back to the application and infrastructure or cloud service where they are hosted. By leveraging graph database technology, CSPs can drive innovation, bring new services from prototype to production, and differentiate their offerings in a competitive market.

This week in Neo4j has seen the release of APOC version 0.8, which includes a new Zeppelin interpreter that connects to Neo4j and allows users to query and display graph data directly in notebooks. Max De Marzi has also been sharing his expertise on building a dating website using Neo4j, while Lasse Westh-Nielsen has shown how to load European road data into Neo4j for path finding queries. Additionally, Andrea Santurbano's Zeppelin interpreter was released as part of the 0.8 release, and he has written a blog post explaining how to build a graph data pipeline using Neo4j and Apache Zeppelin notebooks. The community has also been discussing machine learning and knowledge graphs, with discussions on Twitter about using Neo4j as a master data solution for machine learning systems. The week's featured community member is Andrea Santurbano, who has been contributing to the Neo4j community for several years and has released his Zeppelin interpreter.

SpecterOps, a cybersecurity firm, uses Neo4j to turn attack graphs into defense graphs. They started with BloodHound, a project based on years of work in situational awareness in Active Directory environments. The team chose Neo4j for its simplicity, great documentation, and community support, which made it easy to learn and develop with. However, they found that creating an effective defense graph was surprisingly difficult due to the resilience of attack paths. If they could start over, they would focus on reading more documentation and learning about query efficiency, such as the difference between Shortest Path and All Pairs Shortest Path. The future of graph technology holds opportunities for applying graphs to solving complex cyber security problems, including exploit research, local privilege escalation, and automation of attack path execution.

A knowledge graph is a data graph combined with iterative machine learning, used to solve many enterprise challenges. It's a visual representation of how data is connected and how things are connected. Graphs have a high degree of semantic fidelity, making it easy for business leaders to understand the schema of the graph. They're useful for describing processes and can be used anywhere information needs to flow. Data lakes are never fully completed, but graphs can mobilize data, providing an extensible platform for actionable, end-to-end customer, process, and business analytics. Knowledge graphs provide a transformative platform for solving complex problems, with applications in customer 360 views, real-time recommendation engines, marketing attribution, enterprise search, and proactive risk management. They're also useful for building smart searches, recognizing graph problems, and optimizing data storage.

In the rapidly evolving telecommunications and network operations landscape, service assurance practices have traditionally relied on fragmented views of the network and services, often resulting in inaccurate information. To remain competitive, communication service providers (CSPs) and enterprise IT operations must adopt a holistic, real-time view of their infrastructure and network topology to optimize network services and improve customer satisfaction. Leveraging graph database technology, such as Neo4j, is essential for delivering scalable, nimble solutions that provide a complete view of the network, enabling real-time decision-making and reducing costly service quality issues. By adopting next-generation service assurance (NGSA) solutions with innovations in visibility, scalability, and adaptability, CSPs and IT teams can capitalize on new market opportunities and create unique products and services that uncover hidden patterns and insights, ultimately driving long-term profitability.

This week in Neo4j highlights the latest developments and use cases for graph databases, including the introduction of Neo4j Morpheus, a tool for weaving together graph and relational data in Apache Spark. The community is also showcasing various projects, such as Estelle Joubert's Opera and Musical Canon project using Neo4j to visualize relationships between people and operatic objects. Additionally, there are tutorials on how to use Neo4j with Google Cloud, profiling procedures with JMH, and exploring the World Cup with Neo4j Bloom. The community is also celebrating the release of version 0.8 of Apache Zeppelin, which includes Neo4j support. Furthermore, Dr Jim Webber explains data structures and algorithms used by Neo4j, and there are updates on Neode, a generic OGM built on top of official drivers. The team from APOC is also releasing videos on how to load JSON and JDBC data into Neo4j. Overall, the community is actively exploring new use cases and features for graph databases.

Ajinkya Kale, Senior Applied Researcher at eBay's New Product Development Group, uses Neo4j as a probabilistic graph model to drive conversations in their virtual shopping assistant, eBay ShopBot. He has been using Neo4j since his master's project and credits the platform with helping him troubleshoot issues. The choice of Neo4j was influenced by its track record, ease of use, and ability to visualize data. Ajinkya envisions a future where graph technology is used to encode human knowledge and make decisions, which he believes will be a significant development in the field.

Apache Mesos is a master-agent architecture that optimizes compute resources in the best possible way, using a two-level scheduling system and dealing with resource negotiation, sharing resources across a cluster. DC/OS is a platform on top of Apache Mesos, including Marathon for container orchestration, that adds benefits to run all data services, streaming, machine learning, complex data, and so on. Neo4j Causal Clustering technology provides two parts: core servers, which are in-sync and receive read and write requests, and read replica servers, which are caches but intelligent ones. The clustering approach is essential for overcoming load peaks challenges, and Mesos helps to dynamically shift resources between nodes based on available resources. DC/OS offers local persistent volumes, which provide a place on the disk carrying a label reserved for Neo4j, allowing for restarts of replacement containers exactly on the same data again without full-cluster replication. The combination of Mesos and DC/OS enables running Neo4j Causal Clusters on top of the platform, making it easy to operate and scale the database, along with other microservices like streaming and big data.