General search rules for the Japanese language

In the base configuration, Guidewire provides specialized search algorithms specifically for the Japanese language. Guidewire sets these rules in collations.xml, as described at the beginning of this topic. This Java class provides the following behavior for searching in a Japanese-language database:

Search case	Rule
Half-width/Full-width	All searches in Japanese ignore the difference between half-width and full-width Japanese characters.
Small/Large characters	All searches in Japanese in which the `strength` attribute on `<LinguisticSearchCollation>` is set to `primary`, meaning accent-insensitive, ignore Japanese small/large letter differences in Katakana or Hiragana. Searches in which this attribute is set to `secondary` take small/large letter differences into account.
Katakana and Hiragana	All searches in Japanese ignore the difference between Katakana and Hiragana characters. This type of search is known as kana-insensitive searching.
Long dash (—)	All searches in Japanese ignore the long dash character.
Sound marks ( `` and °)	All searches in Japanese in which the `strength` attribute on `<LinguisticSearchCollation>` is set to `primary` ignore sound marks. Searches in which this attribute is set to `secondary` take sound marks into account.

If you modify the contents of collations.xml or the embedded Java class, PolicyCenter forces a database upgrade the next time the application server starts.