This file lists noteworthy changes between releases, for full list of changes,
see git log and then ChangeLog.old.
- Fixed some twol rules w.r.t. new features that blocked compiling
- Removed lots of dead code
- Autogenerate lexicons from csv data all the time
- Moved to git and googlecode -> chopped most of the documentation and such
- Fixed scripts a bit, added man pages
- Made very crude tests to have at least something back in.
- whole new finntreebank tagset for forthcoming finntreebank work
- uppercasing is noted in the analysis level
- the word boundaries of lexicalised compounds may be available for more cases (depending on the tagset)
- whole new lemmatizer tagset is available
- some dozens of new words added and fixed
- combine corpus analysis script with apertium's preprocessors
- causative derivation chain added
- bbreviations, adpositions, prefixes and suffixes are no longer pos but subcat analyses
- Include deverbal nouns in compounding system
- Start marking compound and strong morpheme boundaries
- New lexical data handling systems
- Implement generator from analyser
- Subcategorize lots of classes for CG and apertium
- Write documentation in booklet format
- New URI and digit string guessers
- New tagging style colorterm for interactive use
- Include weighting scheme in default build
- Demote SUFFIX from POS reading to SUBCAT
- Added marginal enclitics kA, kAs
- Added LEMMA= structure
- re-organized source code to modules
- Added tagging schemes, weighting schemes and suggestion algorithms
- completely new morphology built on traditional lexc-twolc model
- easier route to add new lexical data via simple CSV format
- lots of new lexical data from Joukahainen project as well as extended from kotus-sanalista semi-automatically and by hand.
- titlecasing filter for regular words
- š filter for old orthography variants
- compounding much less haphazard concoction
- parts of speech classified and included
- pronouns, interjections, numerals, proper nouns
- much closer to real full fledged morphology
- movement from SFST to HFST toolset with lots of new cool toys (SFST support is retained in HFST)
- towards full-scale automatic test suite