cdec-dtrain - Mirror of https://github.com/pks/cdec-dtrain.git

Age	Commit message (Collapse)	Author
2013-01-24	Scored grammars from online extraction. Don't trust them yet.	Michael Denkowski

2013-01-07	Online rule extractor output diffs w/ sa extractor	Michael Denkowski

2013-01-07	code cleanup	Michael Denkowski

2013-01-04	Track source span to keep accurate phrase counts	Michael Denkowski

2013-01-04	Fixed issue with overlapping alignment links	Michael Denkowski

2013-01-03	Michael remembers how hiero phrase extraction works. Not totally	Michael Denkowski
	debugged, use -o with caution, etc.
2012-12-28	Collect/store stats from new training instances	Michael Denkowski

2012-12-27	Online phrase extraction speaks rulefactory's language.	Michael Denkowski

2012-12-27	Hooks for online grammar extraction	Michael Denkowski

2012-12-13	Enable loose phrase extraction parameter	Victor Chahuneau
	(default is still tight) use --loose when compiling corpus or tight_phrases = False in config
2012-09-06	add FeatureContext.input_span	Adam Lopez

2012-09-06	[cdec.sa] Allow sentence annotation and initial configuration	Victor Chahuneau

2012-09-06	[cdec.sa] Fix the list of matching training source sentence	Victor Chahuneau

2012-09-06	Merge	Adam Lopez

2012-09-05	Revert to the "old style" pair count...	Victor Chahuneau
	+ API naming fixes + Multiple feature definition files can be passed to the extractor
2012-09-05	Pass F, E texts to features	Adam Lopez

2012-09-05	Change FeatureContext.input_span to return slice indices	Adam Lopez

2012-09-05	Fix bug in initialization of FeatureContext.input_span	Adam Lopez

2012-09-05	Expose new feature extraction API	Victor Chahuneau

2012-09-05	Merge alopez/context-features	Victor Chahuneau

2012-09-03	Support Python 2.6	Victor Chahuneau

2012-08-14	[cdec.sa] Explicit feature names in grammar extractor output	Victor Chahuneau
	+ sparse features in extractor + hg.intersect(string) + basestring = str\|unicode
2012-07-27	[python] conversion from cdec.sa.Rule to cdec.TRule	Victor Chahuneau
	+ remove configobj dependency + re-structure packages (no more top-level library) + "const" stuff + use __new__ instead of constructor for some objects
2012-07-27	[python] Fork of the suffix-array extractor with surface improvements	Victor Chahuneau
	Available as the cdec.sa module, with commande-line helpers: python -m cdec.sa.compile -f ... -e ... -a ... -o sa-out/ -c extract.ini python -m cdec.sa.extract -c extract.ini -g grammars-out/ < input.txt > input.sgml + renamed cdec.scfg -> cdec.sa + Python README