Datasets:
carmentano
commited on
Commit
•
435dd2c
1
Parent(s):
a35c742
Update README.md
Browse files
README.md
CHANGED
@@ -17,7 +17,7 @@ task_categories: []
|
|
17 |
task_ids: []
|
18 |
---
|
19 |
|
20 |
-
#
|
21 |
|
22 |
## Dataset Description
|
23 |
|
@@ -28,11 +28,12 @@ task_ids: []
|
|
28 |
|
29 |
### Dataset Summary
|
30 |
|
31 |
-
This is a dataset for Named Entity Recognition (NER) in Catalan. It adapts <a href="http://clic.ub.edu/corpus/">
|
32 |
|
33 |
-
AnCora corpus is used under [CC-by](https://creativecommons.org/licenses/by/4.0/) licence.
|
|
|
|
|
34 |
|
35 |
-
This dataset was developed by BSC TeMU as part of the AINA project, and to enrich the Catalan Language Understanding Benchmark (CLUB).
|
36 |
|
37 |
### Supported Tasks and Leaderboards
|
38 |
|
@@ -72,7 +73,7 @@ Every file has two columns, with the word form or punctuation symbol in the firs
|
|
72 |
|
73 |
### Data Splits
|
74 |
|
75 |
-
We took the original train, dev and test splits from the [UD version of the corpus]
|
76 |
|
77 |
- train: 10,630 examples
|
78 |
- validation: 1,429 examples
|
@@ -83,7 +84,7 @@ We took the original train, dev and test splits from the [UD version of the corp
|
|
83 |
|
84 |
### Methodology
|
85 |
|
86 |
-
We adapted the NER labels from
|
87 |
|
88 |
### Curation Rationale
|
89 |
|
@@ -92,22 +93,23 @@ We adapted the NER labels from Ancora corpus to a word-per-line format.
|
|
92 |
|
93 |
#### Initial Data Collection and Normalization
|
94 |
|
95 |
-
AnCora consists of a Catalan corpus (AnCora-CA) and a Spanish corpus (AnCora-ES), each of them of 500,000 tokens (some multi-word). The corpora are annotated for linguistic phenomena at different levels.
|
96 |
AnCora corpus is mainly based on newswire texts. For more information, refer to Taulé, M., M.A. Martí, M. Recasens (2009). <a href="http://www.lrec-conf.org/proceedings/lrec2008/pdf/35_paper.pdf">"AnCora: Multilevel Annotated Corpora for Catalan and Spanish”, Proceedings of 6th International Conference on language Resources and Evaluation"</a>.
|
97 |
|
98 |
#### Who are the source language producers?
|
99 |
|
100 |
-
Catalan
|
101 |
|
102 |
### Annotations
|
103 |
|
104 |
#### Annotation process
|
105 |
|
106 |
-
We adapted the NER labels from
|
107 |
|
108 |
#### Who are the annotators?
|
109 |
|
110 |
-
Original annotators from
|
|
|
111 |
### Personal and Sensitive Information
|
112 |
|
113 |
No personal or sensitive information included.
|
|
|
17 |
task_ids: []
|
18 |
---
|
19 |
|
20 |
+
# AnCora-Ca-NER
|
21 |
|
22 |
## Dataset Description
|
23 |
|
|
|
28 |
|
29 |
### Dataset Summary
|
30 |
|
31 |
+
This is a dataset for Named Entity Recognition (NER) in Catalan. It adapts <a href="http://clic.ub.edu/corpus/">AnCora corpus</a> for Machine Learning and Language Model evaluation purposes.
|
32 |
|
33 |
+
[AnCora corpus](http://clic.ub.edu/corpus/) is used under [CC-by](https://creativecommons.org/licenses/by/4.0/) licence.
|
34 |
+
|
35 |
+
This dataset was developed by [BSC TeMU](https://temu.bsc.es/) as part of the [projecte Aina](https://politiquesdigitals.gencat.cat/ca/tic/aina-el-projecte-per-garantir-el-catala-en-lera-digital/), to enrich the [Catalan Language Understanding Benchmark (CLUB)](https://temu-bsc.github.io/catalan-language-understanding-benchmark/).
|
36 |
|
|
|
37 |
|
38 |
### Supported Tasks and Leaderboards
|
39 |
|
|
|
73 |
|
74 |
### Data Splits
|
75 |
|
76 |
+
We took the original train, dev and test splits from the [UD version of the corpus](https://huggingface.co/datasets/universal_dependencies)
|
77 |
|
78 |
- train: 10,630 examples
|
79 |
- validation: 1,429 examples
|
|
|
84 |
|
85 |
### Methodology
|
86 |
|
87 |
+
We adapted the NER labels from [AnCora corpus](http://clic.ub.edu/corpus/) to a word-per-line format.
|
88 |
|
89 |
### Curation Rationale
|
90 |
|
|
|
93 |
|
94 |
#### Initial Data Collection and Normalization
|
95 |
|
96 |
+
[AnCora](http://clic.ub.edu/corpus/) consists of a Catalan corpus (AnCora-CA) and a Spanish corpus (AnCora-ES), each of them of 500,000 tokens (some multi-word). The corpora are annotated for linguistic phenomena at different levels.
|
97 |
AnCora corpus is mainly based on newswire texts. For more information, refer to Taulé, M., M.A. Martí, M. Recasens (2009). <a href="http://www.lrec-conf.org/proceedings/lrec2008/pdf/35_paper.pdf">"AnCora: Multilevel Annotated Corpora for Catalan and Spanish”, Proceedings of 6th International Conference on language Resources and Evaluation"</a>.
|
98 |
|
99 |
#### Who are the source language producers?
|
100 |
|
101 |
+
Catalan [AnCora corpus](http://clic.ub.edu/corpus/) is compiled from articles from the following news outlets: <a href="https://www.efe.com">EFE</a>, <a href="https://www.acn.cat">ACN</a>, <a href="https://www.elperiodico.cat/ca/">El Periodico</a>.
|
102 |
|
103 |
### Annotations
|
104 |
|
105 |
#### Annotation process
|
106 |
|
107 |
+
We adapted the NER labels from [AnCora corpus](http://clic.ub.edu/corpus/) to a token-per-line, multi-column format.
|
108 |
|
109 |
#### Who are the annotators?
|
110 |
|
111 |
+
Original annotators from [AnCora corpus](http://clic.ub.edu/corpus/).
|
112 |
+
|
113 |
### Personal and Sensitive Information
|
114 |
|
115 |
No personal or sensitive information included.
|