Commit Graph

78 Commits

Author SHA1 Message Date
Matthew Honnibal eab2376547 * Allow longer ellipses to be treated as a single token, e.g. Hello......there 2016-05-09 13:22:53 +02:00
Matthew Honnibal 8569dbc2d0 * Add initial stuff for Chinese parsing 2016-04-24 18:44:24 +02:00
Matthew Honnibal fe9299a118 * Fix long-standing issue with coarse-grained tags: proper nouns weren't receiving the PROPN tag, and personal pronouns weren't receiving the PRON tag. This should fix Issue #191, and also Issue #325, which reported that proper nouns were being lemmatized using the common noun policies. This lemmatization will be prevented if the universal tag is PROPN, not NOUN, as no lemmatization rules are loaded for the PROPN tag. 2016-04-14 12:46:43 +02:00
Matthew Honnibal 6f82065761 * Fix infixed commas in tokenizer, re Issue #326. Need to benchmark on empirical data, to make sure this doesn't break other cases. 2016-04-14 11:36:03 +02:00
Matthew Honnibal 85485f5c2b Fix inconsistencies in generate_specials.py
Re Issue #321, fix inconsistencies in the script that generates specials.json. The result still isn't so satisfying --- we need to revise this as we move to parse more morphologically rich languages.
2016-04-07 11:21:52 +10:00
Matthew Honnibal 910a6c805f * Add infix rule for double hyphens, re Issue #302 2016-03-29 13:03:44 +11:00
Wolfgang Seeker eae35e9b27 add tokenizer files for German, add/change code to train German pos tagger
- add files to specify rules for German tokenization
- change generate_specials.py to generate from an external file (abbrev.de.tab)
- copy gazetteer.json from lang_data/en/

- init_model.py
	- change doc freq threshold to 0
- add train_german_tagger.py
	- expects conll09-formatted input
2016-02-18 13:24:20 +01:00
Matthew Honnibal 7cbff48ace * Set the German lemma rules to be an empty JSON object 2016-02-02 22:30:51 +01:00
Matthew Honnibal d0f06c5cc4 * Add missing tags to the German tag map 2016-02-02 22:30:22 +01:00
Matthew Honnibal 6c633f2edc Fix Issue #243: Incorrect gazetteer entry 2016-01-30 06:58:29 +11:00
Matthew Honnibal 4b4eec8b47 * Fix Issue #201: Tokenization of there'll 2015-12-29 18:09:09 +01:00
Matthew Honnibal e8bd92f1e7 * Fix lemma of let's, re Issue #177 2015-11-13 06:42:23 +11:00
Matthew Honnibal 726bb648da * Fix non-breaking space in specials.json 2015-10-19 12:46:11 +11:00
Matthew Honnibal e39095da82 * Fix designation of non-breaking space in specials.json. 2015-10-19 12:39:03 +11:00
Matthew Honnibal 454c1996d0 * Add tokenizer rule to fix numeric range tokenization 2015-10-17 15:49:51 +11:00
Matthew Honnibal 7488821677 * Map NIL to empty string in tag map 2015-10-10 22:09:50 +11:00
Matthew Honnibal 4bbd1388bd * Whitespace 2015-10-10 16:03:48 +11:00
Matthew Honnibal bdcb8d695c * Add non-breaking space to specials.json 2015-10-10 15:54:06 +11:00
Matthew Honnibal c12d36d5f4 * Fix quote marks in lemma_rules 2015-10-10 15:03:36 +11:00
Matthew Honnibal 57b3cd4661 * Add smart-quotes to lemma rules 2015-10-10 14:06:46 +11:00
Matthew Honnibal 7e7f28e1fd * Add smart-quote possessive marker in generate_specials 2015-10-10 14:06:09 +11:00
Matthew Honnibal a510858f5a * Pretty-print specials.json, and add the em dash 2015-10-09 11:07:45 +02:00
Matthew Honnibal 49600a44a8 * Fix trailing comma in lemma_rules.json 2015-10-09 11:06:57 +02:00
Matthew Honnibal 0e92e8574a * Fix pos tag in em-dash in specials 2015-10-09 11:06:37 +02:00
Matthew Honnibal d341443282 * Remove em-dash from lemma rules. Handle instead in specials. 2015-10-09 10:27:13 +02:00
Matthew Honnibal b6047afe4c * Fix punctuation lemma rules, to resolve Issue #130 2015-10-09 10:25:37 +02:00
Matthew Honnibal 393a13d1af * Add unicode em dash to specials.json, so that we can control what POS tag it gets. This way we can prevent sentence boundary detection errors, to address Issue #130. 2015-10-09 19:24:33 +11:00
Matthew Honnibal 1490feda29 * Make generate_specials pretty-print the specials.json file 2015-10-09 19:23:47 +11:00
Matthew Honnibal 1842a53e73 * Lemmatize smart quotes as plain quotes 2015-10-09 19:09:36 +11:00
Matthew Honnibal 5332c0b697 * Add support for punctuation lemmatization, to handle unicode characters. This should help in addressing Issue #130 2015-10-09 18:54:40 +11:00
Matthew Honnibal e3e8994368 * Patch italian tag map 2015-10-08 14:00:13 +11:00
Matthew Honnibal 2d68f75b6a * Fix identity tag map 2015-10-08 13:59:56 +11:00
Matthew Honnibal 095831e5bf * Start adding auxiliaries to morphs.json 2015-09-27 16:56:34 +10:00
Matthew Honnibal c579b6b96c * Update English morphs.json 2015-09-24 22:38:41 +10:00
Matthew Honnibal 3b3547251c * Fix Issue #102: DT tag was mapped to DET. 2015-09-24 18:38:47 +10:00
Matthew Honnibal be4848fbcb * Update morphs.json with universal dependencies/interset morphological features 2015-09-24 00:59:42 +10:00
Henning Peters 911de2ae49 add overseen (?) char 2015-09-22 12:29:47 +02:00
Henning Peters 9ecb98f30e basic german rules 2015-09-22 11:56:29 +02:00
Matthew Honnibal b9e31dc245 * Bug fix to gazetteer.json 2015-09-10 14:50:44 +02:00
Matthew Honnibal 623329b19a Merge branch 'master' of ssh://github.com/honnibal/spaCy into develop 2015-09-08 14:27:01 +02:00
Matthew Honnibal 86c888667f * Merge in changes from de branch 2015-09-06 19:49:28 +02:00
Matthew Honnibal dbf8dce109 Merge branch 'gaz' of ssh://github.com/honnibal/spaCy into gaz 2015-09-06 18:44:14 +02:00
Matthew Honnibal 577418986a * Add draft Italian stuff 2015-09-06 18:44:10 +02:00
Matthew Honnibal 80a66c0159 * Add draft finnish stuff 2015-09-06 18:43:44 +02:00
Matthew Honnibal b3703836f9 * Add en lemma rules 2015-09-06 17:56:11 +02:00
Matthew Honnibal 238b2f533b * Add lemma rules 2015-09-06 17:55:53 +02:00
Matthew Honnibal c9f2082e3c * Fix compilation error in en/tag_map.json 2015-09-06 17:54:51 +02:00
Matthew Honnibal 0af139e183 * Tagger training now working. Still need to test load/save of model. Morphology still broken. 2015-08-27 09:16:11 +02:00
Matthew Honnibal 56c4e07a59 Update gazetteer.json 2015-08-27 08:53:48 +10:00
Matthew Honnibal 494da25872 * Refactor for more universal spacy 2015-08-26 19:13:50 +02:00