cdec-dtrain/training, branch word-alignment Mirror of https://github.com/pks/cdec-dtrain.git https://git.simianer.de/mirrored/cdec-dtrain/atom?h=word-alignment 2010-02-19T03:34:17Z check in modified ones too 2010-02-19T03:34:17Z Chris Dyer redpony@gmail.com 2010-02-19T03:34:17Z urn:sha1:3a7bca942d838f945c1cd0cbe5977e20c61ebc2d add generative word alignment model and primitive EM trainer. Model 1 and HMM are supported, without NULL source words 2010-02-18T22:06:59Z Chris Dyer redpony@gmail.com 2010-02-18T22:06:59Z urn:sha1:4d47dbd7da0434de67ac619392d516c678e1f2ca word aligner cleanup, new features 2010-02-01T22:38:39Z Chris Dyer redpony@gmail.com 2010-02-01T22:38:39Z urn:sha1:c97b8a8b58f7385fb48b74e2cf1ea9610cd1202f fix min 2010-01-25T18:44:58Z Chris Dyer redpony@gmail.com 2010-01-25T18:44:58Z urn:sha1:9e7d60da4421074d279a91cb6e4e67438add4645 more autoconf fixes- use version of boost m4 macros which are much, much, much better 2010-01-24T09:28:43Z Chris Dyer redpony@gmail.com 2010-01-24T09:28:43Z urn:sha1:9670fcbc15419c04f9a03d2118c3c697ed1423a8 Support building without gtest 2010-01-24T08:36:21Z Chris Dyer redpony@gmail.com 2010-01-24T08:36:21Z urn:sha1:613ccc6ea2401e76099a97d3698f13b8ea283497 Now the only dependence is boost, which most modern linux distros have. add alignment visualization tool 2010-01-19T02:57:23Z Chris Dyer redpony@gmail.com 2010-01-19T02:57:23Z urn:sha1:a216521744cd4bf1c9935d99c5e53e4198301b57 cool new alignment feature 2009-12-19T19:32:28Z Chris Dyer redpony@gmail.com 2009-12-19T19:32:28Z urn:sha1:27db9d8c05188f64c17d61c394d3dafe8b8e93d8 add symmetrization heuristics to atools, add null word configuration 2009-12-19T03:51:11Z Chris Dyer redpony@gmail.com 2009-12-19T03:51:11Z urn:sha1:544da4d8e42858b19e6229936df56d44d61b1f38 added non-pruning intersection and a CRF tagger 2009-12-17T18:57:54Z Chris Dyer redpony@gmail.com 2009-12-17T18:57:54Z urn:sha1:bba4ff830c8722cdcaf29e36c1ff5821a912ae5d - the linear-chain tagger is more of a proof of concept than a real tagger-- the context-free assumptions made in a number of places mean that the algorithms used may not be as efficient as they could be, but the model is as powerful as any CRF - it would be easy to add latent variables or semi-CRF support (or both!) - i've added a couple basic features that are often used for POS tagging - non-pruning intersection is useful for lexical word alignment models and the tagger - a sample POS tagger model will be committed later