-
Notifications
You must be signed in to change notification settings - Fork 67
Changelog
Jannik Strötgen edited this page Oct 7, 2016
·
17 revisions
SHA: f7e4c3f
- added: temponym tagging functionality
[Core] - added: English temponym resources
[Resources] - fixed: parameter pos set to "no" (POSTagger.NO) works for all languages now (AllLanguagesTokenizer)
[Standalone] - fixed: several minor issues
SHA: 378e476
- fixed: TreeTaggerWrapper no longer creates temporary files which speeds up processing
[Core] - fixed: Various improvements to Maven support
[Maven] - fixed: Errors in Arabic resources
[Resources] - fixed: Values in IntervalTagger that were switched for a while
[Core]
SHA: b9248e0
- added: automatically-created resources for 200+ languages
[Resources] - added: AllLanguagesTokenizer, a simple, generic, whitespace-based tokenizer that can be used with all languages
- fixed: Minor rule improvements for some languages
[Resources]
SHA: d50e129
- added: Support for Estonian
[Resources] - added: Support for Portuguese (thanks to Zunsik Lim)
[Resources] - added: Resource loading is now easier and and looks in multiple places
[Core] - fixed: StanfordPOSTaggerWrapper would not accept URLs as model (#26)
[Core] - fixed: A bug where pattern replacing in rules would mangle some patterns
[Core] - fixed: Lots of improvements to German and English resources
[Resources] - fixed: Minor improvements for almost all other languages
[Resources]
SHA: b9c5832
- fixed: Italian resources have received a major overhaul
[Resources] - added: Support for Croatian (thanks to Luka Skukan), including a wrapper for the hunpos preprocessing tool
[Resources] - added: Ability to use regular expressions in POS constraints of rules
[Resources] - added: Tokenization without the Tree Tagger's Perl script
[TreeTaggerWrapper] - fixed: various minor bugfixes in the TreeTaggerWrapper and Standalone code
- fixed: Some code pertaining to the invocation of Processors
[Core]
SHA: 5ca451f
- added: Support for calculation of BC and AD dates, and dates close to the year 0, including Arabic, Dutch, English, French, German, Italian, Spanish, Vietnamese language resources.
- added: A preliminary version of Elena Klyachko's Russian resources
[Resources] - fixed: A minor issue with parameter files in TreeTaggerWrapper
[Core]
SHA: d592d19
- added: Chinese resources, support in TreeTaggerWrapper as well as the TempEval-2 Reader and Standalone version
- added: Better handling of overlapping temporal expressions
[Core] - fixed: Made TempEval-3 Reader more robust to non-TE3-inputs
- fixed: More stable TreeTaggerWrapper parameter file recognition
- fixed: Various minor improvements in the resources for all languages
[Resources] - fixed: Some minor fixes for resource recognition in Standalone
[Standalone]
SHA: 183643e
- added: French resources, kindly provided by Véronique Moriceau of the LIMSI-CNRS
[Resources] - added: Support for the IntervalTagger in HeidelTime Standalone
[Standalone] - added: Support to choose from !StanfordPOSTagger or TreeTagger as HeidelTime Standalone's preprocessing engine
[Standalone] - added: Interval resources to Vietnamese
[Resources] - fixed: Improvements in German, English and Vietnamese resources
[Resources] - added: Rudimentary Maven support
[Meta]
SHA: 07c89a5
- fixed: A bug that would prevent HeidelTime Standalone from loading resources under Windows
[Standalone]
SHA: 07c89a5
- added: Support for Spanish and Arabic document processing via standalone
[Standalone] - added: Some more error handling for unexpected user input
[Core] - fixed: Made fixes and alterations to Spanish, Italian, German, Vietnamese and Arabic resources
[Resources] - fixed: A bug where underspecified centuries ("UNDEF-century") in resources would break processing
[Core] - fixed: Made several improvements to the StanfordPOSTagger to work with more unconventional documents
- fixed: Erroneous normalization of underspecified centuries; it now works according to the TIMEX standard
[Resources] - fixed: The Windows
printResourceInformation.batscript now works with paths that contain spaces[Resources]
SHA: 1d2fdfa
- added: Resources for Spanish, Italian, Vietnamese and Arabic
[Resources] - added: TreeTaggerWrapper, a sentence-tokenization, word-tokenization and part of speech tagging wrapper for the popular TreeTagger. It replaces the DKPro Analysis Engines as preprocessing component.
- added: Support for regular expressions in normalization resources
[Core] - added: A new annotator, IntervalTagger that recognizes interval expressions
- added: New UIMA Collection Reader and Consumer for our participation in the TempEval-3 challenge: TempEval3Reader and TempEval3Writer
- added: A UIMA Analysis Engine (JVnTextProWrapper) that leverages the JVnTextPro tool to produce word- and sentence-tokenization as well as part of speech tagging for Vietnamese
- added: A switch (-c) that allows passing the path of the config.props file (issue 3 (on Google Code))
[Standalone] - added: Sub-processor priorities to influence when they are being run
[Core] - added: Switches (-v/-vv) to control the verbosity of logging messages (issue 4 (on Google Code))
[Standalone] - added: A descriptor parameter as well as command line switch (-locale) to specify a locale for HeidelTime to base relative date calculation on (issue 1 (on Google Code))
[Core/Standalone] - added: setenv and printResourceInformation batch files for Windows
- added: An optional setting in the ACETernWriter descriptor to prevent conversion from Timex3 to Timex2
- fixed: Charset in Dutch resources
[Resources] - fixed: Behaviour when non-hardcoded languages are supplied: The resource folder is assumed to be the name of the language
[Core] - fixed: Token boundary detection
[Core] - fixed: Typos in english resources
[Resources] - fixed: A bug where two overlapping temporal expressions would break XML-conformity (issue 5 (on Google Code))
[Standalone] - fixed: A bug that would break resource loading on the Mac OS platform
[Core] - fixed: TempEval2 Reader now works properly with the italian TempEval2 corpus
- fixed: ACE Tern reader recognizes DCTs from the ICAB corpus
- fixed: A bug in TempEval2Writer that would break the output of parallelized UIMA workflows
- fixed: A bug where when you used the HeidelTime standalone version programmatically, the OutputType would be ignored (issue 7 (on Google Code))
[Standalone] - fixed: Several things in a lot of places that made it hard to use HeidelTimeStandalone programmatically (i.a. issue 6 (on Google Code))
[Standalone]
SHA: f1b7a4f
- added: links to the journal paper to the readme file
- fixed: TIMEX3 SET expressions not being translated to TIMEX2 expresions correctly
[ACETernWriter] - removed: dead code
SHA: f814184
- added: support code and english resources for two additional domains: COLLOQUIAL and SCIENTIFIC
- fixed: an incorrect DCT recognition regex
- fixed: standalone not recognizing parameters when not all upper case
SHA: 7dc7ab5
- conducted major code overhaul, mostly modularization-based refactoring
- merged the previously independent heideltime-standalone project into the kit repository
- fixed: made logger message source more apparent
SHA: 38cad56
- added: support to read the ACE Tern 2005 Corpus
- fixed: regex recognitions of strings in the form of "1995-1996"
- added: recognition logic for holidays together with english and german resources
SHA: 8a1262f
This release represents the state of HeidelTime's development before the use of a revision control system; i.e., the version, which was available from the dbs.ifi.uni-heidelberg.de page.