Commit Graph

  • fc75210941 * Move spacy.syntax.conll to spacy.gold Matthew Honnibal 2015-05-24 21:35:02 +0200
  • 765b61cac4 * Update spacy.scorer, to use P/R/F to support tokenization errors Matthew Honnibal 2015-05-24 20:07:18 +0200
  • efe7a7d7d6 * Clean unused functions from spacy.syntax.conll Matthew Honnibal 2015-05-24 20:06:46 +0200
  • 78487f3e66 * Update parser oracle for missing heads Matthew Honnibal 2015-05-24 20:05:58 +0200
  • 541c62c126 * Remove import of removed read_docparse_file function Matthew Honnibal 2015-05-24 20:05:13 +0200
  • 1044a13413 * Begin refactoring scorer to use recall over gold dependencies Matthew Honnibal 2015-05-24 17:40:15 +0200
  • acd1245ad4 * Remove cruft from conll.pyx --- unused stuff about evlauation, which now lives in spacy.scorer Matthew Honnibal 2015-05-24 17:35:49 +0200
  • bfeb29ebd1 * Tmp commit Matthew Honnibal 2015-05-24 02:50:14 +0200
  • 20f1d868a3 * Tmp commit. Working on whole document parsing Matthew Honnibal 2015-05-24 02:49:56 +0200
  • 983d954ef4 * Tmp commit, while switch to new format that assumes alignment happens during training Matthew Honnibal 2015-05-23 17:39:04 +0200
  • f35503018e * Tmp commit of train, while I move to better alignment in gold standard Matthew Honnibal 2015-05-23 17:21:25 +0200
  • bdaddc4103 * Add PTB file read tests Matthew Honnibal 2015-05-20 17:29:27 +0200
  • f2ee9c4feb * Comment out constituency parsing stuff, so that code compiles Matthew Honnibal 2015-05-20 16:55:05 +0200
  • 8ee7c541f1 * Update Constituent definition Matthew Honnibal 2015-05-20 16:03:26 +0200
  • 9dfc9c039c * Work on constituency parsing. Matthew Honnibal 2015-05-20 16:02:51 +0200
  • 8e7d5fce90 * Add members to Tokens documentation Matthew Honnibal 2015-05-13 21:53:41 +0200
  • 5a5710e711 * Fix Span.subtree property Matthew Honnibal 2015-05-13 21:53:15 +0200
  • badf030b6c * Add parse navigation to Span objects Matthew Honnibal 2015-05-13 21:45:19 +0200
  • ca320afe86 * Add docstring for ents attribute Matthew Honnibal 2015-05-13 21:20:47 +0200
  • 7c8bf0eba5 * Add example JSON-formatted training file Matthew Honnibal 2015-05-12 22:42:37 +0200
  • 4230467947 * Update fabfile.py for JSON-formatted training Matthew Honnibal 2015-05-12 22:34:07 +0200
  • ba07b925a7 * Fix compile error in conll.pyx Matthew Honnibal 2015-05-12 22:33:47 +0200
  • f1e0272b18 * Disable c-parsing transitions Matthew Honnibal 2015-05-12 22:33:25 +0200
  • 03a6626545 * Tmp commit Matthew Honnibal 2015-05-11 16:12:03 +0200
  • 9568ebed08 * Fix off-by-one in head reading Matthew Honnibal 2015-05-07 22:53:08 +0200
  • 3d6b3fc6fb * Restore shuffling, and remove print statements from train.py Matthew Honnibal 2015-05-07 22:52:27 +0200
  • e167355505 * Use JSON docs for training and evaluation. Currently a bug that is costing 0.6 acc Matthew Honnibal 2015-05-06 16:38:54 +0200
  • 69840d8cc3 * Tweak verbose output printing in scorer.py Matthew Honnibal 2015-05-06 16:31:23 +0200
  • e0ef6b6992 * Fix alignment in prepare_treebank Matthew Honnibal 2015-05-06 16:31:00 +0200
  • 0605af6838 * Fix head misalignment in read_conll, when periods are ignored Matthew Honnibal 2015-05-06 16:30:28 +0200
  • d2ac8d8007 * Add ctnt field to State, in preparation for constituency parsing Matthew Honnibal 2015-05-06 16:29:10 +0200
  • ab67693393 * Add read_json_file to conll.pyx Matthew Honnibal 2015-05-06 16:27:31 +0200
  • aff9359a8d * Update ner.pyx to expect brackets from gold_tuples Matthew Honnibal 2015-05-06 16:27:01 +0200
  • 0ad72a77ce * Write JSON files, with both dependency and PSG parses Matthew Honnibal 2015-05-05 02:31:20 +0200
  • 5078a32213 * Work on script to format training data as a JSON file. Matthew Honnibal 2015-05-05 01:00:27 +0200
  • d48218f4b2 * Add left_edge and right_edge properties Matthew Honnibal 2015-04-29 22:14:43 +0200
  • bdb56497b5 * Add test for right_edge and left_edge Matthew Honnibal 2015-04-29 22:08:27 +0200
  • 53cf77e1c8 * Bug fix: when non-monotonically correct a dependency, make sure to delete the old one from the child list Matthew Honnibal 2015-04-29 21:32:18 +0200
  • a4e2af54f9 * Add support for l/r edge to add_dep, and move inlined methods into _state.pyx where possible Matthew Honnibal 2015-04-29 19:28:21 +0200
  • d634038eb6 * Add l_edge and r_edge props in TokenC for tracking the parse-yield of the token Matthew Honnibal 2015-04-29 19:14:20 +0200
  • ec42b06a8d * Add download data warning to index.rst Matthew Honnibal 2015-05-12 03:40:58 +0200
  • 194f080b7d * Update index.rst with release blurb Matthew Honnibal 2015-05-12 03:36:57 +0200
  • 3a1ab85a76 * Tweak readme.md Matthew Honnibal 2015-05-12 03:32:53 +0200
  • b9da330d8e * Fix updates description for v0.84 Matthew Honnibal 2015-05-12 03:30:56 +0200
  • 6944a512b7 * Fix updates.rst Matthew Honnibal 2015-05-12 03:27:12 +0200
  • 03ebf70a66 * Inc version to 0.84 Matthew Honnibal 2015-05-12 02:38:51 +0200
  • e73eaf2d05 * Replace some assertions with proper errors Matthew Honnibal 2015-05-08 16:52:17 +0200
  • fb8d50b3d5 Merge branch 'master' of ssh://github.com/honnibal/spaCy Matthew Honnibal 2015-04-30 12:45:15 +0200
  • 4489d87550 * Add cluster=0 by default in init_model Matthew Honnibal 2015-04-29 14:23:13 +0200
  • ed8e8c3bd0 * Whitespace Matthew Honnibal 2015-04-29 14:22:47 +0200
  • 378c2a6435 * Fix POS model: make it use tag instead of pos in history features Matthew Honnibal 2015-04-29 00:02:53 +0200
  • 763ef01575 * Fix two bugs in feature calculation Matthew Honnibal 2015-04-28 23:25:09 +0200
  • 918b820472 * Add testing file for issues such as raised in #57 Matthew Honnibal 2015-04-28 20:46:29 +0200
  • b3fd48c97b * Fix missing root labels bug identified in Issue #57 Matthew Honnibal 2015-04-28 20:45:51 +0200
  • fd71ed5361 Merge pull request #55 from suchow/master Matthew Honnibal 2015-04-21 02:14:47 +1000
  • 3005c86682 Don't track generated data files Jordan Suchow 2015-04-19 13:25:42 -0700
  • 38ed265b7d Tweak line spacing Jordan Suchow 2015-04-19 12:39:18 -0700
  • 85603f5b6a Add CLA for suchow Jordan Suchow 2015-04-19 01:59:42 -0700
  • 1b79d947b9 Minor copyediting Jordan Suchow 2015-04-19 01:56:32 -0700
  • 7bddd15e27 Use consistent sentence spacing within files Jordan Suchow 2015-04-19 01:43:46 -0700
  • 3a8d9b37a6 Remove trailing whitespace Jordan Suchow 2015-04-19 01:31:31 -0700
  • 5f0f940a1f Remove unused imports Jordan Suchow 2015-04-19 01:05:22 -0700
  • 693c5a1558 * Exclude clusterings for words only seen 1 or 2 times, as their clusters are unreliable Matthew Honnibal 2015-04-17 04:44:52 +0200
  • cc4e395927 * Add some ad hoc regexes, for multi-word location prepositions Matthew Honnibal 2015-04-17 04:44:24 +0200
  • f7ffd94e6a * Add Token.conjuncts property Matthew Honnibal 2015-04-17 01:40:53 +0200
  • 4757899370 * Fix times test Matthew Honnibal 2015-04-16 04:50:40 +0200
  • 684d0e5e85 * Download updated data Matthew Honnibal 2015-04-16 04:29:15 +0200
  • 716ba06711 * Inc version Matthew Honnibal 2015-04-16 04:28:15 +0200
  • 2ef170a991 * Fix Issue #54: Error merging multi-word token when there's a mid-token match. Matthew Honnibal 2015-04-16 04:28:06 +0200
  • 42617548af * Disable merge_mwes by default Matthew Honnibal 2015-04-16 04:20:31 +0200
  • 99dbf8a38c * Fix error type in lookup_transition Matthew Honnibal 2015-04-16 01:36:22 +0200
  • 77d0700caf * Add on X way regexes Matthew Honnibal 2015-04-16 01:35:46 +0200
  • adcad4f353 * Clean up train.py Matthew Honnibal 2015-04-15 06:02:04 +0200
  • 9f16848b60 * Add (N0w, N1w) unigram pair to NER features, prompted by failure to detect 'this weekend' Matthew Honnibal 2015-04-15 06:01:18 +0200
  • 05d0f078bb * Inc version Matthew Honnibal 2015-04-13 22:29:31 +0200
  • c6707778dd * Fix Issue #51: Handle non-ascii lemmas correctly Matthew Honnibal 2015-04-13 22:28:59 +0200
  • bf0aff5124 * Fix bug in Tokens.ents where entity wasn't being emitted if another started immediately after Matthew Honnibal 2015-04-13 21:34:33 +0200
  • 0a62c32c14 * Add test for simple NER case Matthew Honnibal 2015-04-13 21:33:54 +0200
  • 49c2e72112 * Upd README Matthew Honnibal 2015-04-13 06:23:30 +0200
  • ab53855dfe * Bump version Matthew Honnibal 2015-04-13 06:08:22 +0200
  • 2b84a90bbb * Fix Issue #50: Python 3 compatibility of v0.80 Matthew Honnibal 2015-04-13 05:59:43 +0200
  • ee317446bd * Try not downloading stuff for travis.yml Matthew Honnibal 2015-04-13 05:59:05 +0200
  • f5d2102421 * Add note to index.rst releasing version 0.80 Matthew Honnibal 2015-04-13 05:45:18 +0200
  • fbd48c571d * Rearrange code in tokens.pyx Matthew Honnibal 2015-04-13 05:41:25 +0200
  • 5ce51ce8d6 * Update docs for v0.80 Matthew Honnibal 2015-04-13 05:40:51 +0200
  • 3faaad0271 * Remove bad import from test_span Matthew Honnibal 2015-04-12 22:42:14 +0200
  • 507048dc45 * Rename StandardError to Exception, for Python 3 compatibility Matthew Honnibal 2015-04-12 07:28:34 +0200
  • c50b493c85 * Add pathlib requirement Matthew Honnibal 2015-04-12 07:17:48 +0200
  • 11c4794e56 * Bump version number Matthew Honnibal 2015-04-12 07:17:32 +0200
  • 761a19113a * Fix /tmp moving thing in download.py Matthew Honnibal 2015-04-12 07:04:10 +0200
  • ed1907b4df * Add pragmatic sentence boundary detection tests, from that Ruby gem. Not automatically run, as they can arbitrarily fail based on model changes. Currently 8/15 fail. Matthew Honnibal 2015-04-12 04:46:40 +0200
  • 0c25001325 * Fix specials.json Matthew Honnibal 2015-04-12 04:45:41 +0200
  • 1629b33082 * Fix copying of tokenizer data in init_model Matthew Honnibal 2015-04-12 04:45:31 +0200
  • 248a2b4b0f * Remove Spans class Matthew Honnibal 2015-04-12 04:07:29 +0200
  • 1d05e6da00 * Add ne_iob and ne_type features to NER Matthew Honnibal 2015-04-10 19:07:08 +0200
  • 4df8a3d90f * Add ne_iob and ne_type attributes to context vector Matthew Honnibal 2015-04-10 05:02:15 +0200
  • 8c354c432b * Add ValueError condition to ner_tag reading Matthew Honnibal 2015-04-10 04:59:59 +0200
  • 435cccf098 * Add read_conll03_file function to conll.pyx Matthew Honnibal 2015-04-10 04:59:11 +0200
  • 99c9ecfc18 * Fix bug in prefix, suffix and word shape features in parser and NER Matthew Honnibal 2015-04-10 03:53:33 +0200
  • a6ac92f077 * Respect the model_dir input parameter to train.py Matthew Honnibal 2015-04-08 22:48:26 +0200