Language and localization

Overview

We manage the display of multi-language data across English and French display locales by carefully leveraging reconciliation and source data language expression:

  • Object classifications are reconciled across source institutions and given DRP-specific labels in English (usually derived from the AAT) and French (usually manually translated) and then inserted at template build time.
  • Language-of-cataloguing is mapped to each object's linked data and then in the presentation layer fields (label, description, etc) are passed to the appropriate language-specific properties in the search index (eg, description.fr vs description.en).
  • Search queries are analyzed in the user's chosen language and then fields are weighted according to that language. For example when French is the chosen language, description.fr is boosted 9 while description.en is boosted 3 to prevent mis-matches of the locale-stemmed query against an index stemmed in a different language, while retaining the ability to match against names and entities within the running text of a field.

References

  • TBD