Thesaurus (Information Retrieval)

Definition

Thesaurus (Information Retrieval)

A thesaurus is a structured vocabulary that records semantic relationships between terms, including synonymy, to support indexing and retrieval. It may also record spelling variants and preferred forms.

Retrieval Uses

• Shared index entry: a normalisation function maps equivalent forms to a common index entry.
• Query expansion: entries remain separate, but a query searches for related alternatives as well as its original terms. Expansion can also be asymmetric.

Grouping forms does not by itself disambiguate homonyms: the appropriate alternatives may depend on the intended meaning.

Example

Synonyms

Car and automobile can share an index label:

Alternatively, keep both entries and expand an automobile query to search for car too.

Spelling variants

Color and colour can map to the same preferred form, despite their different spellings.