The unsegmented nature of Chinese previously caused some typeahead requests to lack precision and/or recall. Typeahead now uses a mixture of linguistic and individual character pattern matching to optimize the balance between precision and recall. Language processing (such as stemming) takes effect for identified words while remaining characters are wildcard-matched. This advantage is made possible through dynamic segmentation.

This enhancement can be enabled through two new parameters, Dx=typeahead+true or Ntx=typeahead+true, or the corresponding Assembler API method. The Ntx=typeahead+true and Dx=typeahead+true parameters indicate that a query should be treated as a typeahead query without the need to specify wildcards. The new interface works with both Latin-1 and OLT language analysis.