Building Knowledge Graphs

Chapter 7. Graph-Native Machine Learning

In this chapter, you’ll learn about the intersection of graphs and ML. You’ll see how ML techniques can automatically enrich an existing knowledge graph as well as how to mine features from a knowledge graph to create accurate predictive models.

This chapter builds on the skills you learned in Chapter 6, where you exploited the topology of a knowledge graph by using graph algorithms. In doing so, you discovered useful insights like the shortest paths between nodes and revealed the communities within the knowledge graph. The same skills, using Neo4j Graph Data Science, Cypher, and Python, will again be called to action in this chapter. You will build on those skills and learn how to add graph-native machine learning to your toolbox to create models that can enrich your knowledge graph.

Like Chapters 3, 5, and 6, this chapter is not intended to be a comprehensive guide to graph-based ML. But it gives you enough information to begin to use knowledge graphs as the basis for ML, with sufficient detail to enable you to explore further should you choose.

Machine Learning in a Nutshell

ML is a huge area of research and practice. But at its most abstract level, it is about deriving programs from data, and of course knowledge graphs are excellent data.

Traditionally, most software has been written to take some input and apply a function to produce some output data. The function is written by an expert human, often with a deep understanding of the ...