Datasets:
Tasks:
Token Classification
Formats:
parquet
Sub-tasks:
named-entity-recognition
Languages:
Arabic
Size:
10K - 100K
License:
Committed local README.md changes before pull.
Browse files
README.md
CHANGED
@@ -1,58 +1,33 @@
|
|
1 |
---
|
2 |
pretty_name: ShamNER
|
3 |
license: cc-by-4.0
|
4 |
-
task_categories:
|
5 |
-
- token-classification
|
6 |
language:
|
7 |
- ar
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
data_files:
|
9 |
train: train.parquet
|
10 |
validation: validation.parquet
|
11 |
test: test.parquet
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
dtype: string
|
26 |
-
- name: text
|
27 |
-
dtype: string
|
28 |
-
- name: source_type
|
29 |
-
dtype: string
|
30 |
-
- name: spans
|
31 |
-
list:
|
32 |
-
- name: annotator
|
33 |
-
dtype: string
|
34 |
-
- name: end
|
35 |
-
dtype: int64
|
36 |
-
- name: label
|
37 |
-
dtype: string
|
38 |
-
- name: start
|
39 |
-
dtype: int64
|
40 |
-
- name: text
|
41 |
-
dtype: string
|
42 |
-
splits:
|
43 |
-
- name: train
|
44 |
-
num_bytes: 5148727
|
45 |
-
num_examples: 19783
|
46 |
-
- name: validation
|
47 |
-
num_bytes: 328887
|
48 |
-
num_examples: 1795
|
49 |
-
- name: test
|
50 |
-
num_bytes: 313228
|
51 |
-
num_examples: 1844
|
52 |
-
download_size: 2302809
|
53 |
-
dataset_size: 5790842
|
54 |
---
|
55 |
-
|
56 |
# ShamNER – Spoken Arabic Named‑Entity Recognition Corpus (Levantine v1.1)
|
57 |
|
58 |
ShamNER is a curated corpus of Levantine‑Arabic sentences annotated for Named Entities, plus dual annotation to check for consisetency (`agreement`) across human annotators.
|
@@ -162,4 +137,4 @@ Every sentence that appears in iaa_A.jsonl is also in the train split (with the
|
|
162 |
|
163 |
`iaa_A.jsonl` and `iaa_B.jsonl` contain parallel annotations for the same 5 806 sentences. Use them to measure agreement or experiment with noise‑robust training. These sentences **do not** overlap with the primary train/val/test splits. As stated above, only `iaa_A.jsonl` were injected into the train, dev and test set.
|
164 |
|
165 |
-
© 2025 · CC BY‑4.0
|
|
|
1 |
---
|
2 |
pretty_name: ShamNER
|
3 |
license: cc-by-4.0
|
|
|
|
|
4 |
language:
|
5 |
- ar
|
6 |
+
annotations_creators:
|
7 |
+
- expert-annotated
|
8 |
+
task_categories:
|
9 |
+
- token-classification
|
10 |
+
task_ids:
|
11 |
+
- named-entity-recognition
|
12 |
+
multilinguality: monolingual
|
13 |
data_files:
|
14 |
train: train.parquet
|
15 |
validation: validation.parquet
|
16 |
test: test.parquet
|
17 |
+
iaa_A: iaa_A.parquet
|
18 |
+
iaa_B: iaa_B.parquet
|
19 |
+
splits:
|
20 |
+
- name: train
|
21 |
+
num_examples: 19532
|
22 |
+
- name: validation
|
23 |
+
num_examples: 1931
|
24 |
+
- name: test
|
25 |
+
num_examples: 1959
|
26 |
+
- name: iaa_A
|
27 |
+
num_examples: 5806
|
28 |
+
- name: iaa_B
|
29 |
+
num_examples: 5806
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
30 |
---
|
|
|
31 |
# ShamNER – Spoken Arabic Named‑Entity Recognition Corpus (Levantine v1.1)
|
32 |
|
33 |
ShamNER is a curated corpus of Levantine‑Arabic sentences annotated for Named Entities, plus dual annotation to check for consisetency (`agreement`) across human annotators.
|
|
|
137 |
|
138 |
`iaa_A.jsonl` and `iaa_B.jsonl` contain parallel annotations for the same 5 806 sentences. Use them to measure agreement or experiment with noise‑robust training. These sentences **do not** overlap with the primary train/val/test splits. As stated above, only `iaa_A.jsonl` were injected into the train, dev and test set.
|
139 |
|
140 |
+
© 2025 · CC BY‑4.0
|