Celebrate the Success Stories of Knowledge Management - 2022 KMWorld Awards

Graph databases team up with BI: It’s all about relationships

Article Featured Image

Connecting with ease

Seeing the value of integrating graph databases with traditional BI solutions, and also responding to customer demand, Neo4j introduced its BI Connector in March 2020. “Analysts wanted to use our database, but they did not have technical backgrounds to carry out analyses and needed a familiar interface,” said Amy Hodler, director of graph analytics and AI
programs at Neo4j. “With the BI Connector, they can consume the analyses that the Neo4j graph database provides with
the tools they are accustomed to using.”

The BI Connector translates SQL queries into Cypher, Neo4j’s query language. The SQL queries can be generated in Tableau, Qlik, Looker, Oracle Analytics Cloud, TIBCO Spotfire, and MicroStrategy Secure Analytics. These products use Java Database Connectivity (JDBC), an API for Java. A connector is also in the works for Open Database Connectivity (ODBC), which is used by Microsoft’s BI solution, PowerBI, and many other products. The results of the queries carried out by Cypher are then presented in the BI tool deployed in the enterprise. Depending on how the BI solution is designed, the system can be a view-only or interactive dashboard.

IBM developed a dashboard to support business process improvement for the U.S. Department of Agriculture (USDA) using Neo4j’s graph database and BI Connector. The processes that the USDA was analyzing were converted to graphs that reflected relationships among the elements that comprise the organizational processes, such as stakeholders, meetings, and activity lifecycles. Queries input via a Tableau interface were converted to the Cypher query language. The results
were delivered to a Tableau server that presented them on a dashboard to the enduser. Tableau brings data from different
sources together, and provides visualization of the data.

Neo4j recognizes that BI software is widely used and deeply embedded in many enterprises, but it does not address all analytical needs. For fraud detection, BI solutions can be enhanced by graph databases that focus on connections. “When you look at fraud through a graph lens, there are certain activities that have unique profiles, such as identity theft versus group fraud,” Hodler observed. “These ‘islands’ of interaction may not reflect large dollar amounts individually but they can be an indicator of a significant problem.”

Another use case for combining traditional BI with graph data analytics is in financial services. “If a bank is looking for a high-value customer in order to upsell, the usual approach would be to identify individuals with large accounts,” noted Hodler. However, it is possible through the use of graph analytics to discover that someone with a small account is associated with a group of people who are CEOs. “Graph analytics can reveal indirect connections,” she continued, “or unusually high interaction levels, or behaviors in common. That connection might identify them as a potential high-value individual even if their account at that institution is small.” Banking analysts can combine existing customer information with an influence or value score from Neo4j based on the graph algorithm and support a decision about which customer to
contact and what offer to make.

New possibilities

Despite the potential offered by integrating graph databases with traditional BI solutions, the use of this combination has been limited so far. The biggest obstacle is not technology, but mindset, according to Morris. “Most BI solutions are legacy applications with well-defined functions,” he said. “Users have some very specific deliverables that they need and expect. It is difficult to persuade an enterprise that it may not be using the right platform, or that it needs a different vision for the future. However, if they are going to be true learning organizations and stay agile in a changing world, they need to be open to new possibilities.” With the increasing awareness of graph databases and the way in which they complement
other databases, more organizations may decide to tackle this challenge.

KMWorld Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues
Companies and Suppliers Mentioned