“We want to help the world make sense of data and we think graphs are the best way of doing that.”
That’s the word from Emil Eifrem, CEO of Neo Technology, which makes the open-source Neo4j NoSQL graph database. He’s not talking in terms of RDF-centric solutions, even though he says he’s 100 percent in agreement with the vision of the semantic web and machine readability. “The world is a graph,” Eifrem says, “and RDF is a great way of connecting things. I’m all in agreement there.” The problem, in his opinion, is that execution on the software end there has been lacking.
“This comes down to usability,” he says, and the average developer, he believes, finds the semantic web-oriented tools largely incomprehensible. Eifrem says he’s speaking from real-world experiences, having worked directly with RDF and taught classes on the semantic web layers. Where it took a week to get students up to speed on things like Jena and Sesame, they ‘get’ the property graph and graph databases in half-a-day, he says. Neo4j stores data in nodes connected by directed, typed relationships with properties on both – also known as a property graph.
The property graph, he readily acknowledges, where nodes contain key value pairs, is inferior to the RDF model of graphs that has nodes and labeled directed arcs that link pairs of nodes. The property graph “is much less capable,” he says. That being the case, he argues that it’s the slices removed from the property graph model that are the ones most people have a hard time understanding as they are represented in the semantic web model. “So, it’s less capable but people get it.”
And that is why property graphs make the better connection with his own mission to make the world love graphs. “My goal is to appeal to a mass audience. Graphs are so amusing, amazing and capable they have to be in the hands of everyone. They have to be in the hands of the average developer.”
Eifrem says that appeal is taking root among enterprise developers in companies like VMware, Cisco, Viadeo and Deutsche Telekom, who along with other customers are using Neo4j variously for conducting high performance queries on huge amounts of data related to social, recommendations, bioinformatics, fraud detection, network management, authorization and access control, master data management, content management, and even parcel routing. “Graph databases are already at a point way beyond where the semantic web has even been commercially,” he says. “And also they have the power of a more simple model.”
The range of vertical industries Neo Technology is seeing adopt its solution, he says, speak to the inevitability of having to exploit graphs. “The semantic web guys could have captured this,” he believes, "but the technology is too difficult for normal developers.”
Whether or not you agree, for users that want to know more about graph databases, Eifrem participated with Ian Robinson and Jim Webber on the O’Reilly book Graph Databases, which interested parties can download free here.