Document Selection

Document selection retrieves a manageable candidate set before detailed ranking. Follow the example query through an inverted index, then combine term matching, popularity, intent, and personalization into an initial relevance score.

This lesson examines the document-selection stage of the funnel introduced in the previous lesson:

For this example, assume a corpus of one-hundred billion documents and a target of one-hundred thousand candidates per query. Information retrieval supplies the initial matching and scoring methods. These are sizing assumptions for the design, not a measurement of the current web.

Information retrieval is the science of searching for information in a document. It focuses on comparing the query text with the document text and determining what is a good match.

Documents

Document types are as follows: