According to Max De Marzi, software field engineer with Neo Technology, graph searches offer real value to the enterprise by making practical use of big data. There are insights and patterns to be uncovered simply by being able to see how things are connected. De Marzi uses Neo4j, Neo Technology's graph database, and Cypher, the Neo4j query language, to build proof of concepts for professional customers. In other words, he demonstrates how this technology can be applied and what business problems it could solve.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
His more recent proof of concept was inspired by Facebook's Graph Search, a search engine providing user-specific search results based on "natural language" queries. At this year's Big Data TechCon, de Marzi will teach a class on how to build your own Facebook Graph Search using your own big data.
There is already a wealth of information to be tapped into. All that's needed is a dynamic way to synthesize and apply it.
Putting big data to work
Industries that most directly benefit from big data are customer-facing ones -- namely, retail. "I logged in to the retailer website and it asked me what my general interests were, and then that's as far as they go," De Marzi said. He went on to postulate that if it had access to data on his family, his kids and his wife, and if it had identified their ages and the things they liked, the website would have been able to precisely target products that might interest him.
"And the thing is, it's about leverage. You can leverage social data. You can leverage Twitter data. You don't have to start from scratch." There is already a wealth of information to be tapped into. All that's needed is a dynamic way to synthesize and apply it.
While retail is the most obvious beneficiary of graph searches, De Marzi cited use cases from a wide variety of industries. For example, banks can use big databases to protect themselves from fraud by detecting suspicious connections. "If some credit card is being used by 20 different people, it may be a fraudulent card," De Marzi explained.
More sophisticated scams, such as fraud rings, rely on burying data under layers of misdirection. Traditional methods would process this information as individual data points and would miss their incriminating interrelatedness -- such as 20 people using the same card.
Making your own graph search
"There's a bit of a learning curve, especially if you're coming from a NoSQL background," De Marzi cautioned. For one thing, developers will have to learn to interpret natural language in order to create Cypher queries. That said, De Marzi believes that the primary challenge is in mind shift. Developers will need to wrap their heads around the concept of data points being dynamic, interconnected objects.
"That usually can take anywhere from a couple of hours to a couple of days just to adjust," De Marzi said. "Once their heads wrap around it, everything starts to make sense and it becomes simpler and easy to use."
The good news is there's plenty of material out there to help developers get started. "Now, there's so much information, so much online training and books. I think people are hitting the ground running."