3
I Use This!
Moderate Activity

Commits : Listings

Analyzed 1 day ago. based on code collected 2 days ago.
May 15, 2023 — May 15, 2024
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
Update to v4.5.7 - improved dependency converter, F1 scores for individual conparse trees More... 28 days ago
Find fronted 'said' ccomps, with a few negative detections to avoid other likely tree structures which are similar but aren't actually fronted ccomp - see comments More... 28 days ago
remove json-simple from pom More... about 1 month ago
Adding PTB Corrector as an option reduces the total validation errors in the PTB conversion to dependencies by about 250. Weirdly this is by removing 280 syntax errors and adding 40 morpho errors for aux verbs. Presumably those should be fixable. Of course, there is always more that can be done - there are now 2622 errors left when using the converter. https://github.com/UniversalDependencies/docs/issues/717 More... about 2 months ago
Add the ability to EnglishPTBTreebankCorrector to use it as a TreeTransformer with a single transformation, rather than on an entire treebank More... about 2 months ago
Not sure how this didn't blow up in the past 10 years... perhaps tsurgeon is more restrictive about not having the labels in the query now More... about 2 months ago
Don't CC underneath an MWE, since those are deliberately expected to be FIXED. Improves 'all but' and 'whether or not' https://github.com/UniversalDependencies/docs/issues/717 More... 2 months ago
Add a no-self-loop rule to the very general VP < VP < other verb AUX rule in UniversalEnglishGrammaticalRelations. This doesn't change any trees in the PTB Train section, but does help the tag updater not hit false positives for the AUX rules https://github.com/UniversalDependencies/docs/issues/717 More... 2 months ago
Technically the UniversalPOSMapper needs to use the UniversalSemanticHeadFinder, in case any of its POS mapping rules use # More... 2 months ago
Make the tsurgeon debug mode noisier More... 2 months ago
Gottem More... 2 months ago
Run a TreeTransformer that gets rid of the functional tags other than TMP on the NPs before using the POS conversions. This also greatly reduces the number of validation errors, especially thanks to the AUX rules now matching for NP-stuff whereas before it would not match those More... 2 months ago
Add a DEBUG flag to the UniversalPOSMapper which outputs the automatically generated tregex/tsurgeons More... 2 months ago
Update the UniversalPOSMapper to use AUX for a large chunk of the dependencies by reusing the patterns from UniversalEnglishGrammaticalRelations to find those words. Currently it is finding more than it should, but the error rate is significantly lower than it is without this change More... 2 months ago
The root of a tregex expression now keeps track of what variables it knows about More... 2 months ago
Split the context mappings into two separate arrays - this gives us a place to reuse the AUX mappings from UniversalEnglishGrammaticalRelations More... 2 months ago
Whitespace More... 2 months ago
Simplify all the 1-1 tag mappings in UniversalPOSMapper More... 2 months ago
Refactor somewhat - move all the context-sensitive mappings to their own construction More... 2 months ago
Initial version: turn the UniversalPOSMapper tsurgeon file into code. This will allow us to easily programmatically reuse the AUX rules, which currently do not match the UD standard for UPOS More... 2 months ago
Tiny test of the UniversalPOSMapper More... 2 months ago
Updated the combiner to mark a couple known lemmas More... 2 months ago
Add lemmas to a few of the MWTs that we combine for English. A few others are still TODO, such as the n't 'll etc suite More... 2 months ago
Treat 'dinna' as an MWT in the converter More... 2 months ago
Fix doc error More... 2 months ago
Oops, the flat needs to include the possible NN tag as part of its target More... 3 months ago
Add a FLAT relation for phrases such as 'en masse' which aren't considered a FIXED or MWT expression. Addresses a tiny part of https://github.com/UniversalDependencies/docs/issues/717 More... 3 months ago
Can treat 'sort of' the same as 'kind of' when converting constituency trees to dependencies. https://github.com/UniversalDependencies/docs/issues/717 More... 3 months ago
Add a sent_id to the UD Converter, since the validation script used in UD tools seems to throw a fit over not having a sent_id More... 3 months ago
Add the f1 scores for each tree to the parser protobuf responses More... 3 months ago