You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@opennlp.apache.org by kr...@apache.org on 2026/09/06 21:58:43 UTC
(opennlp) branch OPENNLP-547-dependency-parser updated (5cc405820 -> afd68fb1f)
This is an automated email from the ASF dual-hosted git repository.
krickert pushed a change to branch OPENNLP-547-dependency-parser
in repository https://gitbox.apache.org/repos/asf/opennlp.git
from 5cc405820 OPENNLP-547: Document the remaining constructors and private contracts
add 49b9612df OPENNLP-1888: Document annotation container: typed offset-anchored layers over the original text
add 3e382b31b OPENNLP-1888: Lemma and stem layer adapters
add e8a2ca47f OPENNLP-1888: Pipeline example and contract tests for the document container, javadoc precision pass
add e9b53a7a6 OPENNLP-1888: Parse per sentence in the adapters, distinguish empty from missing layers, validate providers at build time
add 6b384b670 OPENNLP-1888: Lemmatize per sentence, validate adapter outputs, clear adaptive data on failure
add 99e0b83fe OPENNLP-1888: Add StringUtil.isBlank following the toolkit whitespace definition
add eb277fbb6 OPENNLP-1888: Manual chapter, review-convention pass, and key placement rule
add 1ba2fcdae OPENNLP-1888: Namespace the standard layer key ids
add f8b2ee97e OPENNLP-1888: Declare per-key positional or document scope
add b39dd25d1 OPENNLP-1888: State the index-reference invariants in the specification text
add e4d6ed718 OPENNLP-1888: Document the gold-layer convention
add 266e780d9 OPENNLP-1888: Create toolkit layer keys through namespace-applying factories on Layers
add d43c1ec9a OPENNLP-1888: Javadoc container overrides and cite the pipeline example test
add 04532170c OPENNLP-1888: Anchor the Document thread-safety note to immutability
add 15102d75d OPENNLP-1888: Address review: shared annotator helpers, hardened container contract
add e9062250b OPENNLP-1888: Address review: fold duplicate test fixtures and pin null rejection
add 689636105 OPENNLP-1888: Name the null argument, avoid per-call key set wrapping, pin blank and span edge cases
add 608430990 OPENNLP-1888: Rework the document chapter after docs review
add 9442b91b6 OPENNLP-1888: Pin the Document.merge contract
add 919cf590e OPENNLP-1888: Implement Document.merge as a default method
add aa5aebae7 OPENNLP-1888: Document the parallel fan-out join in the manual
add adb15b426 OPENNLP-1888: Pin the duplicate-layer policy contract for merge
add 1297f725d OPENNLP-1888: Add a duplicate-layer policy to Document.merge
add 3245eff1c OPENNLP-1888: Render the layer figure monospaced and tighten the prose
add 4f5c0cacd OPENNLP-1888: Pin a differing-contents message for KEEP_EQUAL rejection
add 844128fad OPENNLP-1888: Name the differing contents when KEEP_EQUAL rejects a layer
add 75fcac098 OPENNLP-1888: Pin what layer equality compares for KEEP_EQUAL
add 562af8714 OPENNLP-1888: Pin merge semantics across Document implementations
add 09cd47d59 OPENNLP-1888: Merge with a single map copy in ImmutableDocument
add 0705c5d73 OPENNLP-1888: State span offsets in UTF-16 units in the manual
add 550bc4490 OPENNLP-1888: Pin that adapters name themselves by class name
add f4dd343a9 OPENNLP-1888: Name adapters by their simple class name
add e380352c3 OPENNLP-1888: Move concrete annotators to runtime
add 22780da3d OPENNLP-1888: Place annotators with their components
add 5115aaa40 depparse: Arc-standard dependency parser skeleton with CoNLL-U reader and UAS/LAS evaluator
add f5c11dc46 depparse: DependencyModel persistence and a gated Universal Dependencies evaluation
add 1d8c366e2 depparse: Read CoNLL-U raw so multiword-token sentences keep their dependencies
add dc070db0e depparse: Valency and child features for the transition classifier
add 282e886a5 depparse: Pure-Java feedforward neural tier, training and inference in the JVM
add 70987c15d depparse: Pretrained word embedding seeding for the feedforward trainer
add b0a3ed6b3 depparse: Beam search decoding for the feedforward parser
add 2be280921 depparse: Global refinement with beam search and early update
add d1334ad27 depparse: State, transition, usage, and edge-case tests; expanded trainer and parser javadoc
add a70ae3840 depparse: Document treebank acquisition for the gated evaluation with a download helper
add e48a5c948 depparse: Refine into a new model, fail loud on unknown transitions, and read blanks and case through the project seams
add e83f03cb5 depparse: Restore the final-sigma rule in vocabulary normalization and make model bytes reproducible
add c8584a265 depparse: Cache hidden-layer contributions per feature pair on frozen models
add 9eb6c1b6c depparse: Validate relation labels through StringUtil.isBlank
add d894cfc66 dependency: Add a dependency parser manual chapter with a mirror-tested example
add e15e35e47 depparse: Address review: decode outcomes once, fold duplication, name constants
add 62a412dc3 depparse: Address review: cite specs and document private helpers
add 84f059e4d OPENNLP-547: Parse CoNLL-U fields without regex
add ffb99d009 depparse: Arc-standard dependency parser skeleton with CoNLL-U reader and UAS/LAS evaluator
add c828b8666 depparse: DependencyModel persistence and a gated Universal Dependencies evaluation
add f7182f63d depparse: Read CoNLL-U raw so multiword-token sentences keep their dependencies
add 14a226a25 depparse: Valency and child features for the transition classifier
add cf9080ba8 depparse: Pure-Java feedforward neural tier, training and inference in the JVM
add 944cff4d0 depparse: Pretrained word embedding seeding for the feedforward trainer
add ecfdc9eff depparse: Beam search decoding for the feedforward parser
add 63cd289fc depparse: Global refinement with beam search and early update
add 273499018 depparse: State, transition, usage, and edge-case tests; expanded trainer and parser javadoc
add 752c05c5c depparse: Document treebank acquisition for the gated evaluation with a download helper
add d1072b78e depparse: Refine into a new model, fail loud on unknown transitions, and read blanks and case through the project seams
add 09e82da86 depparse: Restore the final-sigma rule in vocabulary normalization and make model bytes reproducible
add 03a005dbc depparse: Cache hidden-layer contributions per feature pair on frozen models
add 299b16f5c depparse: Validate relation labels through StringUtil.isBlank
add 0e13326f0 dependency: Add a dependency parser manual chapter with a mirror-tested example
add ef853a5bb depparse: Address review: decode outcomes once, fold duplication, name constants
add 89825abbc depparse: Address review: cite specs and document private helpers
add e69d29820 OPENNLP-547: Parse CoNLL-U fields without regex
add 4d8ec0290 OPENNLP-547: Reconcile branch after document API cascade
add 65c215b69 OPENNLP-1888: Add a ChunkerAnnotator providing a chunks layer
add 1b24daddd OPENNLP-1888: Add a ParserAnnotator providing a phrases layer
add e37f9c108 OPENNLP-1888: List the chunker and parser adapters in the manual
add 71da20136 OPENNLP-547: Merge updated document container branch
add 0763c2136 OPENNLP-547: Reconcile published parser history
add 535c6d43a Cancel superseded pull request CI runs (#1267)
add d6e74c4f0 Merge current main into dependency-parser
add b81a2a7d1 OPENNLP-1925: Document RegEx exemption and harden ReDoS guard test (#1268)
add 34ff57bcf OPENNLP-1893: Correct Hunspell parsing and analysis (#1266)
add 266b6d419 Merge Apache main through Hunspell follow-up #1266 into OPENNLP-547-dependency-parser
add 9715fff9c OPENNLP-547: Show that an underscore relation is read as a label
add fb55c0fea OPENNLP-547: Show that reserved symbols in training data break reloading
add 40186281b OPENNLP-547: Skip CoNLL-U trees with a placeholder relation
add 549338a7e OPENNLP-547: Keep reserved embedding entries when training data spells them
add 14f029ef8 OPENNLP-547: Evaluate both parsers on the UD English development split
add 7712bf332 OPENNLP-547: Report attachment scores without punctuation
add 5b380e5f9 OPENNLP-547: Add a cross validator for dependency parsers
add 1ba04efd1 OPENNLP-547: Complete Javadoc, validation, and constants in the dependency parser
add 279da7bf0 OPENNLP-547: Parameterize the dependency parser tests and name their fixtures
add f24dd1547 OPENNLP-547: Back every manual example of the dependency parser with a test
add 205359e1e OPENNLP-547: Evaluate the dependency parsers across four UD 2.0 treebanks
add 0893833ee OPENNLP-547: Document what the dependency parser evaluation covers
add 48ec28b3e OPENNLP-547: Expect InvalidFormatException for malformed models and word lines
add 1f4ed1b19 OPENNLP-547: Report format faults as InvalidFormatException
add 0e6511f92 OPENNLP-547: Test projectivity before deriving oracle transitions
add afd68fb1f OPENNLP-547: Keep parser internals package-private and validate input in the module
No new revisions were added by this update.
Summary of changes:
.github/workflows/allowlist-check.yml | 7 +
.github/workflows/maven.yml | 7 +
dev/README-hunspell-dictionaries.md | 19 +-
.../opennlp/tools/depparse/DependencyGraph.java | 33 +-
.../opennlp-formats/dev/README-ud-treebanks.md | 41 +-
.../opennlp-formats/dev/download-ud-treebank.sh | 24 +-
.../conllu/ConlluDependencySampleStream.java | 54 +-
.../conllu/ConlluDependencyParserEvalTest.java | 81 ---
.../conllu/ConlluDependencyParserUsageTest.java | 102 +++-
.../conllu/ConlluDependencySampleStreamTest.java | 35 +-
.../conllu/UdTreebankDownloadScriptTest.java | 4 +
.../opennlp/tools/depparse/ArcStandardOracle.java | 41 +-
.../opennlp/tools/depparse/ArcStandardState.java | 8 +-
.../tools/depparse/DependencyContextGenerator.java | 9 +-
.../tools/depparse/DependencyCrossValidator.java | 186 ++++++
.../tools/depparse/DependencyEvaluator.java | 73 ++-
.../tools/depparse/DependencyEventStream.java | 32 +-
.../opennlp/tools/depparse/DependencyModel.java | 39 +-
.../opennlp/tools/depparse/DependencyParserME.java | 15 +-
.../tools/depparse/FeedforwardDependencyModel.java | 234 ++++++--
.../depparse/FeedforwardDependencyParser.java | 31 +-
.../depparse/FeedforwardDependencyTrainer.java | 246 ++++++--
.../java/opennlp/tools/depparse/ParserInput.java | 59 ++
.../java/opennlp/tools/depparse/Transition.java | 3 +
.../opennlp/tools/namefind/RegexNameFinder.java | 3 +
.../tools/namefind/RegexNameFinderFactory.java | 3 +
.../tools/stemmer/hunspell/HunspellDictionary.java | 635 +++++++++++++++++----
.../tools/stemmer/hunspell/HunspellStemmer.java | 209 +++++--
.../tools/depparse/ArcStandardOracleTest.java | 35 +-
.../depparse/DependencyCrossValidatorTest.java | 197 +++++++
.../tools/depparse/DependencyEvaluatorTest.java | 108 ++++
.../depparse/DependencyParserEdgeCaseTest.java | 97 ++--
.../tools/depparse/DependencyParserMETest.java | 81 ++-
.../tools/depparse/DependencyTestSamples.java | 20 +-
.../depparse/FeedforwardDependencyParserTest.java | 355 +++++++++---
.../tools/namefind/RegexNameFinderFactoryTest.java | 47 +-
.../stemmer/hunspell/HunspellStemmerTest.java | 548 +++++++++++++++++-
opennlp-docs/src/docbkx/dependency.xml | 31 +-
opennlp-docs/src/docbkx/stemmer.xml | 20 +-
.../tools/eval/UniversalDependencyParserEval.java | 352 ++++++++++++
40 files changed, 3487 insertions(+), 637 deletions(-)
delete mode 100644 opennlp-core/opennlp-formats/src/test/java/opennlp/tools/formats/conllu/ConlluDependencyParserEvalTest.java
create mode 100644 opennlp-core/opennlp-runtime/src/main/java/opennlp/tools/depparse/DependencyCrossValidator.java
create mode 100644 opennlp-core/opennlp-runtime/src/main/java/opennlp/tools/depparse/ParserInput.java
create mode 100644 opennlp-core/opennlp-runtime/src/test/java/opennlp/tools/depparse/DependencyCrossValidatorTest.java
create mode 100644 opennlp-core/opennlp-runtime/src/test/java/opennlp/tools/depparse/DependencyEvaluatorTest.java
create mode 100644 opennlp-eval-tests/src/test/java/opennlp/tools/eval/UniversalDependencyParserEval.java