Added swedish lemma rules

2024-09-21 11:29:13 +03:00 · 2017-02-04 17:53:32 +01:00 · 2017-02-04 17:53:32 +01:00 · 19c0ce745a
commit 19c0ce745a
parent 69eab727d7
1 changed files with 45 additions and 0 deletions
--- a/spacy/sv/lemma_rules.py
+++ b/spacy/sv/lemma_rules.py
@ -0,0 +1,45 @@
+# encoding: utf8
+from __future__ import unicode_literals
+
+
+LEMMA_RULES = {
+    "noun": [
+        ["t", ""],
+        ["n", ""],
+        ["na", ""],
+        ["na", "e"],
+        ["or", "a"],
+        ["orna", "a"],
+        ["et", ""],
+        ["en", ""],
+        ["en", "e"],
+        ["er", ""],
+        ["erna", ""],
+        ["ar", "e"],
+        ["ar", ""],
+        ["lar", "el"],
+        ["arna", "e"],
+        ["arna", ""],
+        ["larna", "el"]
+    ],
+
+    "adj": [
+        ["are", ""],
+        ["ast", ""],
+        ["re", ""],
+        ["st", ""],
+        ["ägre", "åg"],
+        ["ägst", "åg"],
+        ["ängre", "ång"],
+        ["ängst", "ång"],
+        ["örre", "or"],
+        ["örst", "or"],
+    ],
+
+    "punct": [
+        ["“", "\""],
+        ["”", "\""],
+        ["\u2018", "'"],
+        ["\u2019", "'"]
+    ]
+}