collations

A collation is a set of rules that determine how strings are compared and ordered. Collations reflect linguistic conventions and depend on locale. They govern which characters are considered equal, how accents and case are treated, and how punctuation and digits influence ordering.

Collations define strength or sensitivity levels that describe how strictly differences are treated. At the primary level, differences such as case and diacritics are ignored; secondary and tertiary levels distinguish accents and case; higher levels may also consider punctuation and spacing. Unicode normalization and canonical equivalence influence comparisons across character forms, especially for characters that have multiple valid representations.

Standards include the Unicode Collation Algorithm (UCA), which provides a language-independent framework and uses data from CLDR. Implementations exist in ICU and in many programming libraries; databases expose collations as locale- or charset-specific rules. Binary or raw collations compare bytes directly and are deterministic but do not reflect natural language ordering.

In databases and software, collations affect sorting (ORDER BY), indexing, and string comparisons for equality. They enable locale-aware search and consistent results across applications, but can complicate cross-language data and introduce performance considerations. Before applying a collation, normalization to a canonical Unicode form (for example NFC) is often advisable to ensure stable comparisons.

Common database examples include utf8_general_ci and utf8_unicode_ci in MySQL, locale-based collations in PostgreSQL, COLLATE clauses in SQL Server, and Oracle NLS_SORT options.