Merge pull request #10 from richardpaulhudson/support/3.4.1

Support for v3.4.1 English spaCy models

.github/workflows/test-coreferee.yml

@@ -12,6 +12,10 @@ jobs:
         spacy_version: ['3.3.0']
         click_version: ['8.0.1']
         include:
+          - os: 'ubuntu-latest'
+            python-version: '3.9'
+            spacy_version: '3.4.2'
+            click_version: '8.0.1'
           - os: 'ubuntu-latest'
             python-version: '3.9'
             spacy_version: '3.2.0'

README.md

@@ -31,6 +31,7 @@ Author: <a href="mailto:[email protected]">Richard Paul Hudson, Explosion AI<
   - [6.6 Version 1.1.3](#version-113)
   - [6.7 Version 1.2.0](#version-120)
   - [6.8 Version 1.3.0](#version-130)
+  - [6.9 Version 1.3.1](#version-131)
 - [7. Open issues/requests for assistance](#open-issues)
 
 <a id="introduction"></a>
@@ -43,7 +44,7 @@ Author: <a href="mailto:[email protected]">Richard Paul Hudson, Explosion AI<
 
 Coreferences are situations where two or more words within a text refer to the same entity, e.g. _**John** went home because **he** was tired_. Resolving coreferences is an important general task within the natural language processing field.
 
-Coreferee is a Python 3 library (tested with versions 3.6—3.10) that is used together with [spaCy](https://spacy.io/) (tested with versions 3.0.0—3.4.0) to resolve coreferences within English, French, German and Polish texts. It is designed so that it is easy to add support for new languages. It uses a mixture of neural networks and programmed rules.
+Coreferee is a Python 3 library (tested with versions 3.6—3.10) that is used together with [spaCy](https://spacy.io/) (tested with versions 3.0.0—3.4.2) to resolve coreferences within English, French, German and Polish texts. It is designed so that it is easy to add support for new languages. It uses a mixture of neural networks and programmed rules.
 
 The library was originally developed at [msg systems](https://www.msg.group/en), but is now being maintained at [Explosion AI](https://explosion.ai).
 
@@ -273,7 +274,7 @@ Coreferee started life to assist the [Holmes](https://github.com/explosion/holme
 <table style="text-align:center; vertical-align:middle">
   <tr><td rowspan="2">ISO 639-1</td><td rowspan="2">Language</td><td rowspan="2">Training corpora</td><td rowspan="2">Total words in training corpora</td><td colspan="2"><code>*_trf</code> models</td><td colspan="2"><code>*_lg</code> models</td><td colspan="2"><code>*_md</code> models</td><td colspan="2"><code>*_sm</code> models</td></tr>  
   <tr><td align="center">Anaphors in 20%</td><td align="center">Accuracy (%)</td><td align="center">Anaphors in 20%</td><td align="center">Accuracy (%)</td><td align="center">Anaphors in 20%</td><td align="center">Accuracy (%)</td><td align="center">Anaphors in 20%</td><td align="center">Accuracy (%)</td></tr>
-  <tr><td align="center">en</td><td align="center">English</td><td align="center"><a href="https://opus.nlpl.eu/ParCor/">ParCor</a>/<a href="https://github.com/dbamman/litbank"> LitBank</a></td><td align="center">393564</td><td align="center"><b>2500—2580</b></td><td align="center"><b>80—83</b><td align="center"><b>2480—2520</b></td><td align="center"><b>81—82</b></td></td><td align="center">2480—2510</td><td align="center">81</td><td align="center">2540—2560</td><td align="center">81—82</td></tr>
+  <tr><td align="center">en</td><td align="center">English</td><td align="center"><a href="https://opus.nlpl.eu/ParCor/">ParCor</a>/<a href="https://github.com/dbamman/litbank"> LitBank</a></td><td align="center">393564</td><td align="center"><b>2500—2580</b></td><td align="center"><b>80—83</b><td align="center"><b>2480—2520</b></td><td align="center"><b>81—82</b></td></td><td align="center">2480—2510</td><td align="center">81-83</td><td align="center">2510—2560</td><td align="center">81—82</td></tr>
   <tr><td align="center">de</td><td align="center">German</td><td align="center"><a href="https://opus.nlpl.eu/ParCor/">ParCor</a></td><td align="center">164300</td><td align="center">-</td><td align="center">-</td><td align="center"><b>530—570</b></td><td align="center"><b>79—80</b></td><td align="center">520—550</td><td align="center">79—80</td><td align="center">530—550</td><td align="center">76—79</td></tr>
   <tr><td align="center">fr</td><td align="center">French</td><td align="center"><a href="https://www.ortolang.fr/market/corpora/democrat/v1.1">DEMOCRAT</a></td><td align="center">323754</td><td align="center">-</td><td align="center">-</td><td align="center"><b>1270—1280</b></td><td align="center"><b>71—72</b></td><td align="center">1280—1300</td><td align="center">68—70</td><td align="center">1130—1140</td><td align="center">63—64</td></tr>
   <tr><td align="center">pl</td><td align="center">Polish</td><td align="center"><a href="http://zil.ipipan.waw.pl/PolishCoreferenceCorpus">PCC</a></td><td align="center">548268</td><td align="center">-</td><td align="center">-</td><td align="center"><b>1730—1790</b></td><td align="center"><b>72—76</b></td><td align="center">1740—1790</td><td align="center">70—75</td><td align="center">-</td><td align="center">-</td></tr>
@@ -628,6 +629,12 @@ The initial open-source version.
 
 - Added support for spaCy v3.4 for English, German and Polish.
 
+<a id="version-131"></a>
+
+##### 6.9 Version 1.3.1
+
+- Added support for the v3.4.1 English models.
+
 <a id="open-issues"></a>
 
 ### 7. Open issues / requests for assistance

SHORTREADME.md

@@ -1,6 +1,6 @@
 Coreferences are situations where two or more words within a text refer to the same entity, e.g. _**John** went home because **he** was tired_. Resolving coreferences is an important general task within the natural language processing field.
 
-Coreferee is a Python 3 library (tested with versions 3.6—3.10) that is used together with [spaCy](https://spacy.io/) (tested with versions 3.0.0—3.4.0) to resolve coreferences within English, French, German and Polish texts. It is designed so that it is easy to add support for new languages. It uses a mixture of neural networks and programmed rules.
+Coreferee is a Python 3 library (tested with versions 3.6—3.10) that is used together with [spaCy](https://spacy.io/) (tested with versions 3.0.0—3.4.2) to resolve coreferences within English, French, German and Polish texts. It is designed so that it is easy to add support for new languages. It uses a mixture of neural networks and programmed rules.
 
 The library was originally developed at [msg systems](https://www.msg.group/en), but is now being maintained at [Explosion AI](https://explosion.ai).
 

coreferee/lang/en/config.cfg

@@ -26,25 +26,25 @@ vectors_model:          core_web_lg
 [lg_3_4_0]
 model:                  core_web_lg
 from_version:           3.4.0
-to_version:             3.4.0
+to_version:             3.4.1
 train_version:          3.4.0
 
 [md_3_4_0]
 model:                  core_web_md
 from_version:           3.4.0
-to_version:             3.4.0
+to_version:             3.4.1
 train_version:          3.4.0
 
 [sm_3_4_0]
 model:                  core_web_sm
 from_version:           3.4.0
-to_version:             3.4.0
+to_version:             3.4.1
 train_version:          3.4.0
 
 [trf_3_4_0]
 model:                  core_web_trf
 from_version:           3.4.0
-to_version:             3.4.0
+to_version:             3.4.1
 train_version:          3.4.0
 vectors_model:          core_web_lg
 

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = coreferee
-version = 1.3.0
+version = 1.3.1
 description = Coreference resolution for English, French, German and Polish, optimised for limited training data and easily extensible for further languages
 long_description = file: SHORTREADME.md
 long_description_content_type = text/markdown

tests/en/test_rules_en.py

@@ -435,8 +435,12 @@ def test_potential_pair_he_she_antecedent_person_noun(self):
     @unittest.skipIf(train_version_mismatch, train_version_mismatch_message)
     def test_potential_pair_he_she_antecedent_non_person_proper_noun(self):
         self.compare_potential_pair(
-            "I worked for Skateboards plc. She was there", 4, False, 6, 1,
-            excluded_nlps=["core_web_sm"]
+            "I worked for Skateboards plc. She was there",
+            4,
+            False,
+            6,
+            1,
+            excluded_nlps=["core_web_sm"],
         )
 
     def test_potential_pair_it_exclusively_person_antecedent(self):
@@ -491,6 +495,7 @@ def test_potential_pair_she_exclusively_male_name_compound_antecedent_control(se
     def test_potential_pair_person_word_non_capitalized(self):
         self.compare_potential_pair("I saw a job. He was there", 3, False, 5, 0)
 
+    @unittest.skipIf(train_version_mismatch, train_version_mismatch_message)
     def test_potential_pair_person_word_capitalized(self):
         self.compare_potential_pair("I saw Job. He was there", 2, False, 4, 1)