This chapter covers
- The mechanics of query interpretation
- Implementing an end-to-end query intent pipeline
- Tagging and classifying terms and phrases in queries
- Augmenting queries with knowledge graph traversals
- Interpreting semantics of domain-specific query patterns
In the last two chapters, we covered how to use your content and signals to interpret the domain-specific meaning of incoming user queries. We discussed phrase identification, misspelling detection, synonym discovery, query intent classification, related terms expansion, and even query-sense disambiguation. We’ve mostly discussed these techniques in isolation, however, to demonstrate how they each work independently.
In this chapter, we’ll put all those learnings into practice, integrating the techniques into an end-to-end query interpretation framework. We’ll show an example search interface that accepts real queries, interprets them, rewrites them to better express the end user’s intent, and then returns ranked results.