This site is not optimized for Internet Explorer 9 and lower. Please choose another browser or upgrade your existing browser in order get the best experience of this website.

Graph Advantage: Real-Time Recommendations

GraphAdvantageRealTimeRecommendationsWe all receive recommendations presented to us on a daily basis. From the products that we should be buying to the movies we should be watching to the people we should be dating…the list goes on. You’re capable of recommending just about anything — as long as you have the right data in place. Graph databases are naturally well-suited for building real-time recommendation engines thanks to the native graph traversal performance when traversing the network around and between the desired starting node such as a person that already bought a set of products.

Whether an enterprise functions within the social, media or retail sector, providing users with targeted, real-time recommendations are important for providing the customer value through a personalized experience, which is quickly becoming the baseline for remaining competitive. Unlike that of business data, recommendations should be contextual and inductive so it can be deemed relevant by the end consumer. Achieving this requires a “good enough” level of data classification with sufficient connectedness between the data points in the system.

With a graph database where relationships are treated as first class citizens, you can connect a customer’s browsing history while combining that with their purchase history and offline product and brand interactions to enable the real-time recommendation algorithm to utilize their present choices and offer personalized recommendations without any offline pre-compute delaying the interaction — lowering the potential for the consumer to purchase from a competitor.

Neo4j for Real-Time Recommendations

Whether you’re leveraging social connections or connecting data across digital and physical customer touch points, the Neo4j graph database provides the possibility of providing relevant real-time recommendations to customers. It also offers a plethora of advantages including:

  • A native graph store: Neo4j houses connected user and purchase data that’s neither hierarchical nor linear. Its native graph storing capability makes it simpler to decode suggestion data without any intermediate indexing at each turn.
  • A versatile schema: The flexible graph model of Neo4j allows for groups to evolve recommendation engines efficiently and try variations simply by traversing different paths through the graph.
  • Scalability and performance: The native graph processing engine of Neo4j is capable of supporting sub-second graph queries on large data sets to allow for real-time decision making even when 10’s of thousands of nodes are involved in the network analysis of the algorithm execution.

Real-Time Data Changes

The beauty of utilizing a connected graph database, like Neo4j, is that it makes real-time recommendations very powerful by the simple fact that as soon as a new relationship is being drawn, results will immediately change because it will be considered in the recommendation pattern. Such an update happens within seconds as opposed to relying on offline processes that need to compute and cache the recommendations or other data systems that can’t connect data by default.

Storing and querying recommendation data via the Neo4j graph database lets your application offer real-time results instead of dealing with pre-calculate data. With customer expectations on the rise, offering real-time recommendations will provide a competitive advantage in the near-term and become essential in the long-term.