Are you able to get a really good look at what the data in your graph database can tell you? A graph database such as Aurelius’ Titan, Franz’ AllegroGraph, Google’s Cayley, Neo Technology’s Neo4j, Objectivity’s InfiniteGraph, and Ontotext’s GraphDB lets you store semi-structured data linked together by relationships in a graph and make inferences from existing facts. On their own, they are incredibly powerful for use cases that involve making and traversing data connections, such as social network interactions, e-commerce recommendations, financial transactions, and mapping network dependencies.
But that power can increase when graph data visualization tools enter the picture to make it easier for business users to search, investigate, and analyze data, exposing patterns and trends within a graph’s content that mere mortals can quickly identify and react to.
“Data visualization matters because it is a technique with which humans can understand data,” says Jean Villedieu, co-founder of Linkurious, which unveiled its Linkurious Enterprise graph data visualization platform for Neo4j this spring. That understanding helps people make critical decisions, whether it’s freezing a bank account after spotting a suspicious pattern in financial interactions across entities that could indicate money laundering, or determining from relationships the next piece of hardware among network components that may be subject to failure and the group of users that could affect, so that network administrators can quickly react to the issue.
Linkurious’ Enterprise software joins other solutions that can propel graph data visualizations, including its own linkurious.js. That is a graph visualization library like D3.js, Keylines, VivaGraph and Sigma.js, he explains. Linkurious’ solution actually leverages Sigma.js for its graph data structure and visualization engine. Linkurious has enhanced Sigma.js beyond its efficiency as a graph viewer for the application development market with higher-lever and integration-ready features, such as filters and Excel exporters, for creating smart graph applications, and with some 20 plug-ins to improve its core with new interaction features. There also is Gephi, an Open Source graph visualization tool that is popular among scientists and Data Scientists. In fact, Linkurious CEO Sebastian Heymann co-created Gephi, which has been downloaded about one million times in its latest incarnation. Neo4J has its own built-in data visualization tool, too, based on the D3.js library for manipulating documents based on data.
“In the market you have visualization libraries, that require team plus resources to integrate, and tools for scientists,” says Villedieu. “The big difference of Linkurious Enterprise is that it’s easy to use and works out of the box.”
Out-of-the-Box and into the Business
While there are some very powerful data visualization products to choose from that help companies interact with data and make decisions based on that, especially at the Data Scientist level, there’s been a gap when it comes to providing an out-of-the-box option for business users working with Big Data that is stored in terms of the connections that exist between people, places, objects, products, and so on, as represented in graph databases. Linkurious Enterprise is there to fill that gap, Villedieu says.
According to “Data Visualization: A HorizonWatch Trend Report,” among the top data visualization trends for this year, based on internal IBM analysis, was “increased interest by LOB for all types of data discovery and visualization tool information, demos and case studies,” as well as the desire by users to “be able [to] manipulate data themselves.” That report wasn’t specific to graph data visualization. But, Villedieu’s own thoughts align with that position, especially as graph databases are gaining interest in the wider enterprise community – beyond specific verticals such as intelligence or specific job profiles such as data analysts – or storing and analyzing connected data on a large scale.
“Visualization is the next step in making graph databases something widespread,” he says. He likens it to what happened a couple of decades ago with relational databases – that is, that these tools were once limited in their reach, but the rise of solutions such as IBM Cognos that enabled data visualization made them more accessible to business people. “The same thing is going to happen for graph databases,” he says. Linkurious wants to be a leader in the space, and to that end it expects in the future to work with other graph databases as well as Neo4J, he notes.
The idea with Linkurious Enterprise, he says, is to democratize graph visualization for companies that need to extract insights from big and complex datasets easily, safely, and collaboratively. According to Villedieu, companies want to be able to extend data visualization capabilities to their user bases, but they also demand enterprise-like features to assure that that data is appropriately accessible without being jeopardized. That’s been a challenge.
The five pillars of the Linkurious Enterprise solution that addresses this – and other issues – are that it:
- Provides an interface to search and explore graphs visually and interactively, using tools like a lasso to select multiple nodes, so that everyone can find answers in graph data, as well as makes it possible to refine queries with Boolean operators or by filtering on multiple properties at once. That said, “the search is also and mostly enabled via a search engine à la Google,” Villedieu says – type text and get results.
- Makes it possible for administrators to control access to data through user LDAP-authentication and access rules based on categories of nodes and relationships. “Administrators can quickly define an authentication scheme for the level of access for different users,” says Villedieu. “Security is definitely one of the most important things about Linkurious Enterprise.”
- Enables automatic connection to the graph database via a Web browser. “If the data already is stored in a Neo4J graph database it’s very straightforward,” he says. “It’s just a matter of minutes to get started.”
- Lets users save graph exploration results as a visualization that they themselves can access from their dashboards, share with colleagues or export to formats like Gephi and Excel.
- Supports opportunities to enrich graphs by modifying, creating or deleting nodes and relationships based on the security rules the organization sets.
Data Visualization in Action
Data visualization works well when companies think very closely about what they want to ask of the data and publish about it, Villedieu thinks, and when they take full advantage of data visualization techniques. For example, they can choose to associate certain data with a certain color, or text or icon size, “and that can help make the visualization more impactful because it conveys more meaning at once,” he says. One color can convey that a particular node is a person while another is a company, for instance, while increasingly larger icons can indicate increasingly larger financial transaction sizes.
Some 200 customers have put its technology to use so far, he says, including Cisco, eBay, and the French Ministry of Finances. Perhaps one of its better known stories is how the International Consortium of Investigative Journalists (ICIJ) used Linkurious as part of a six-month investigation that led to reporting how HSBC, one of the world’s biggest banks, helped over 100,000 customers hide $100 billion in Switzerland. The investigative team transformed leaked financial data from the Swiss branch of HSBC to a Neo4j graph database to identify all the persons and bank accounts in data connected to each other, and who controlled which accounts. The story was revealed in February by more than 50 news organizations worldwide.
“These were journalists investigating this,” Villedieu notes, most of them without Data Science or development experience. “They used Linkurious to reach their conclusions about who was involved in the tax fraud. It was called the biggest data investigation of the decade.”
That, he hopes, is an example of how graph visualization will become possible for all, no matter their level of data expertise – and sooner rather than later.
