Notes from a work meeting with Fran and Sjur

  • list bugs we found, priorities for the work (on the wiki page)
    • bugs: no disambiguation, unrec. word disappears in cg output, blue underline on wrong word (due to no disamb), gramcheck crashing on *any* data outside the expected xml in errors.xml
  • work on 2.0 of the zhfst specification
  • fst-based preprocessor
    • punctuation as a separate loop or separate fst before and after
  • run disambiguation cg using cg3_sentence_copy