Compare direction without rewarding length

A journey through distance and similarity with Botie.

Natural Language Processing · A journey with Botie

First visit? Learn the workshop controls

Botie's notebook

How Could a Search Engine Compare Two Lists of Words?

Part 1 of 3

A query and a document rarely contain exactly the same text. A search system therefore needs a graded comparison rather than an exact match. Once terms are represented as coordinates, the problem becomes deciding which geometric relationship reflects relevance.

Sources for this history

Salton, Wong and Yang (1975), A Vector Space Model for Automatic Indexing

← Previous lesson: A Vector for a Word Read the full lesson and derivations