5 Knowledge graph learning

 

This chapter covers

  • Building and working with knowledge graphs
  • Implementing open information extraction to generate knowledge graphs from text
  • Discovering arbitrary semantic relationships with semantic knowledge graphs
  • Query expansion and rewriting using knowledge graphs
  • Interpreting documents with knowledge graphs

In the last chapter, we primarily focused on learning the similarity between queries and documents based on users’ behavioral signals. In chapter 2, we also discussed how textual document content, instead of being “unstructured data”, is more like a giant graph of hyper-structured data containing a rich graph of semantic relationships connecting the many character sequences, terms, and phrases that exist across our collections of documents.

In this chapter, we’ll demonstrate how to use this giant graph of semantic relationships within our content to better interpret domain-specific terminology. We’ll accomplish this by using both traditional knowledge graphs, which enable explicit modeling of relationships within a domain, and semantic knowledge graphs, which enable real-time inference of nuanced semantic relationships within a domain.

5.1 Working with knowledge graphs

5.2 Using our search engine as a knowledge graph

5.3 Automatically extracting knowledge graphs from content

5.3.1 Extracting arbitrary relationships from text

5.3.2 Extracting hyponyms and hypernyms from text

5.4 Learning intent by traversing semantic knowledge graphs

5.4.1 What is a semantic knowledge graph?

5.4.2 Indexing the datasets

5.4.3 Structure of an SKG

5.4.4 Calculating edge weights to measure the relatedness of nodes

5.4.5 Using SKGs for query expansion

5.4.6 Using SKGs for content-based recommendations

5.4.7 Using SKGs to model arbitrary relationships

5.5 Using knowledge graphs for semantic search

Summary