Datasets:
Tasks:
Text2Text Generation
Sub-tasks:
text-simplification
Languages:
German
Size:
10K<n<100K
ArXiv:
License:
Update README.md
Browse files
README.md
CHANGED
@@ -6,11 +6,14 @@ language:
|
|
6 |
pretty_name: DEplain-APA
|
7 |
size_categories:
|
8 |
- 10K<n<100K
|
|
|
|
|
9 |
---
|
10 |
|
11 |
-
# Dataset
|
|
|
12 |
|
13 |
-
|
14 |
- [Dataset Description](#dataset-description)
|
15 |
- [Dataset Summary](#dataset-summary)
|
16 |
- [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
|
@@ -34,46 +37,71 @@ size_categories:
|
|
34 |
- [Citation Information](#citation-information)
|
35 |
- [Contributions](#contributions)
|
36 |
|
37 |
-
|
38 |
|
39 |
- **Repository:** [DEplain-APA zenodo repository](https://zenodo.org/record/7674560)
|
40 |
- **Paper:** Regina Stodden, Momen Omar, and Laura Kallmeyer. 2023. ["DEplain: A German Parallel Corpus with Intralingual Translations into Plain Language for Sentence and Document Simplification."](https://arxiv.org/abs/2305.18939). In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Toronto, Canada. Association for Computational Linguistics.
|
41 |
- **Point of Contact:** [Regina Stodden](regina.stodden@hhu.de)
|
42 |
|
43 |
-
|
44 |
|
45 |
-
|
46 |
|
47 |
-
|
48 |
|
49 |
The dataset supports the training and evaluation of `text-simplification` systems. Success in this task is typically measured using the [SARI](https://huggingface.co/metrics/sari) and [FKBLEU](https://huggingface.co/metrics/fkbleu) metrics described in the paper [Optimizing Statistical Machine Translation for Text Simplification](https://www.aclweb.org/anthology/Q16-1029.pdf).
|
50 |
|
51 |
-
|
52 |
|
53 |
The text in this dataset is in Austrian German (`de-at`).
|
54 |
|
55 |
-
|
56 |
All texts in this dataset are news data.
|
57 |
|
58 |
## Dataset Structure
|
59 |
|
60 |
-
|
61 |
|
62 |
- The dataset is licensed with restricted access for only academic purposes. To download the dataset, please request access on [zenodo](https://zenodo.org/record/7674560).
|
63 |
|
64 |
-
|
65 |
-
- `document-simplification` configuration: an instance consists of an original document and one reference simplification.
|
66 |
-
- `sentence-simplification` configuration: an instance consists of
|
67 |
-
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
-
|
72 |
-
|
73 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
74 |
|
75 |
|
76 |
-
|
77 |
|
78 |
DEplain-APA is randomly split into a training, development and test set. The training set of the sentence-simplification configuration contains only texts of documents which are part of the training set of document-simplification configuration and the same for dev and test sets.
|
79 |
The statistics are given below.
|
@@ -81,72 +109,88 @@ The statistics are given below.
|
|
81 |
|
82 |
| | Train | Dev | Test | Total |
|
83 |
| ----- | ------ | ------ | ---- | ----- |
|
84 |
-
| Document Pairs |
|
85 |
-
| Sentence Pairs |
|
86 |
|
|
|
87 |
|
88 |
Here, more information on simplification operations will follow soon.
|
89 |
|
90 |
-
|
91 |
|
92 |
-
|
93 |
|
94 |
DEplain-APA was created to improve the training and evaluation of German document and sentence simplification. The data is provided by the same data provided as for the APA-LHA data. In comparison to APA-LHA (automatic-aligned), the sentence pairs of DEplain-APA are all manually aligned. Further, DEplain-APA aligns the texts in language level B1 with the texts in A2, which result in mostly mild simplifications.
|
95 |
|
96 |
-
Further DEplain-APA, contains parallel documents as well as parallel sentence pairs.
|
97 |
|
98 |
-
|
99 |
|
100 |
-
|
101 |
|
102 |
-
The original news texts (in CEFR level
|
103 |
All documents date back to 2019 to 2021.
|
104 |
Two German native speakers have manually aligned the sentence pairs by using the text simplification annotation tool TS-ANNO. The data was split into sentences using a German model of SpaCy.
|
105 |
|
106 |
-
|
107 |
-
The original news texts (in CEFR level
|
108 |
|
109 |
-
|
110 |
|
111 |
-
|
112 |
|
113 |
The instructions given to the annotators are available [here](https://github.com/rstodden/TS_annotation_tool/tree/master/annotation_schema).
|
114 |
|
115 |
-
|
116 |
|
117 |
The annotators are two German native speakers, who are trained in linguistics. Both were at least compensated with the minimum wage of their country of residence.
|
118 |
They are not part of any target group of text simplification.
|
119 |
|
120 |
-
|
121 |
|
122 |
No sensitive data.
|
123 |
|
124 |
-
|
125 |
|
126 |
-
|
127 |
|
128 |
Many people do not understand texts due to their complexity. With automatic text simplification methods, the texts can be simplified for them. Our new training data can benefit in training a TS model.
|
129 |
|
130 |
-
|
131 |
|
132 |
No bias is known.
|
133 |
|
134 |
-
|
135 |
|
136 |
The dataset is provided for research purposes only. Please check the dataset license for additional information.
|
137 |
|
138 |
-
|
139 |
|
140 |
-
|
141 |
|
142 |
Researchers at the Heinrich-Heine-University Düsseldorf, Germany, developed DEplain-APA. This research is part of the PhD-program `Online Participation` supported by the North Rhine-Westphalian (German) funding scheme `Forschungskolleg`.
|
143 |
|
144 |
-
|
|
|
|
|
145 |
|
146 |
-
|
147 |
|
148 |
-
|
149 |
|
150 |
-
[More Information Needed]
|
151 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
152 |
This dataset card uses material written by [Juan Diego Rodriguez](https://github.com/juand-r) and [Yacine Jernite](https://github.com/yjernite).
|
|
|
6 |
pretty_name: DEplain-APA
|
7 |
size_categories:
|
8 |
- 10K<n<100K
|
9 |
+
task_ids:
|
10 |
+
- text-simplification
|
11 |
---
|
12 |
|
13 |
+
# Dataset Statement for DEplain-APA
|
14 |
+
In the following, we provide a dataset for DEplain-APA (following Huggingface's data cards).
|
15 |
|
16 |
+
### Table of Contents
|
17 |
- [Dataset Description](#dataset-description)
|
18 |
- [Dataset Summary](#dataset-summary)
|
19 |
- [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
|
|
|
37 |
- [Citation Information](#citation-information)
|
38 |
- [Contributions](#contributions)
|
39 |
|
40 |
+
### Dataset Description
|
41 |
|
42 |
- **Repository:** [DEplain-APA zenodo repository](https://zenodo.org/record/7674560)
|
43 |
- **Paper:** Regina Stodden, Momen Omar, and Laura Kallmeyer. 2023. ["DEplain: A German Parallel Corpus with Intralingual Translations into Plain Language for Sentence and Document Simplification."](https://arxiv.org/abs/2305.18939). In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Toronto, Canada. Association for Computational Linguistics.
|
44 |
- **Point of Contact:** [Regina Stodden](regina.stodden@hhu.de)
|
45 |
|
46 |
+
#### Dataset Summary
|
47 |
|
48 |
+
DEplain-APA [(Stodden et al., 2023)](https://arxiv.org/abs/2305.18939) is a dataset for the training and evaluation of sentence and document simplification in German. All texts of this dataset are provided by the Austrian Press Agency. The simple-complex sentence pairs are manually aligned.
|
49 |
|
50 |
+
#### Supported Tasks and Leaderboards
|
51 |
|
52 |
The dataset supports the training and evaluation of `text-simplification` systems. Success in this task is typically measured using the [SARI](https://huggingface.co/metrics/sari) and [FKBLEU](https://huggingface.co/metrics/fkbleu) metrics described in the paper [Optimizing Statistical Machine Translation for Text Simplification](https://www.aclweb.org/anthology/Q16-1029.pdf).
|
53 |
|
54 |
+
#### Languages
|
55 |
|
56 |
The text in this dataset is in Austrian German (`de-at`).
|
57 |
|
58 |
+
#### Domains
|
59 |
All texts in this dataset are news data.
|
60 |
|
61 |
## Dataset Structure
|
62 |
|
63 |
+
#### Data Access
|
64 |
|
65 |
- The dataset is licensed with restricted access for only academic purposes. To download the dataset, please request access on [zenodo](https://zenodo.org/record/7674560).
|
66 |
|
67 |
+
#### Data Instances
|
68 |
+
- `document-simplification` configuration: an instance consists of an original document and one reference simplification (in plain-text format).
|
69 |
+
- `sentence-simplification` configuration: an instance consists of original sentence(s) and one manually aligned reference simplification (inclusing one or more sentences).
|
70 |
+
|
71 |
+
|
72 |
+
#### Data Fields
|
73 |
+
|
74 |
+
| data field | data field description |
|
75 |
+
|-------------------------------------------------|-------------------------------------------------------------------------------------------------------|
|
76 |
+
| `original` | an original text from the source dataset |
|
77 |
+
| `simplification` | a simplified text from the source dataset |
|
78 |
+
| `pair_id` | document pair id |
|
79 |
+
| `complex_document_id ` (on doc-level) | id of complex document (-1) |
|
80 |
+
| `simple_document_id ` (on doc-level) | id of simple document (-0) |
|
81 |
+
| `original_id ` (on sent-level) | id of sentence(s) of the original text |
|
82 |
+
| `simplification_id ` (on sent-level) | id of sentence(s) of the simplified text |
|
83 |
+
| `domain ` | text domain of the document pair |
|
84 |
+
| `corpus ` | subcorpus name |
|
85 |
+
| `simple_url ` | origin URL of the simplified document |
|
86 |
+
| `complex_url ` | origin URL of the simplified document |
|
87 |
+
| `simple_level ` or `language_level_simple ` | required CEFR language level to understand the simplified document |
|
88 |
+
| `complex_level ` or `language_level_original ` | required CEFR language level to understand the original document |
|
89 |
+
| `simple_location_html ` | location on hard disk where the HTML file of the simple document is stored |
|
90 |
+
| `complex_location_html ` | location on hard disk where the HTML file of the original document is stored |
|
91 |
+
| `simple_location_txt ` | location on hard disk where the content extracted from the HTML file of the simple document is stored |
|
92 |
+
| `complex_location_txt ` | location on hard disk where the content extracted from the HTML file of the simple document is stored |
|
93 |
+
| `alignment_location ` | location on hard disk where the alignment is stored |
|
94 |
+
| `simple_author ` | author (or copyright owner) of the simplified document |
|
95 |
+
| `complex_author ` | author (or copyright owner) of the original document |
|
96 |
+
| `simple_title ` | title of the simplified document |
|
97 |
+
| `complex_title ` | title of the original document |
|
98 |
+
| `license ` | license of the data |
|
99 |
+
| `last_access ` or `access_date` | data origin data or data when the HTML files were downloaded |
|
100 |
+
| `rater` | id of the rater who annotated the sentence pair |
|
101 |
+
| `alignment` | type of alignment, e.g., 1:1, 1:n, n:1 or n:m |
|
102 |
|
103 |
|
104 |
+
#### Data Splits
|
105 |
|
106 |
DEplain-APA is randomly split into a training, development and test set. The training set of the sentence-simplification configuration contains only texts of documents which are part of the training set of document-simplification configuration and the same for dev and test sets.
|
107 |
The statistics are given below.
|
|
|
109 |
|
110 |
| | Train | Dev | Test | Total |
|
111 |
| ----- | ------ | ------ | ---- | ----- |
|
112 |
+
| Document Pairs | 387 | 48 | 48 |483 |
|
113 |
+
| Sentence Pairs | 10660 | 1231 | 1231 | 13122|
|
114 |
|
115 |
+
Inter-Annotator-Agreement: 0.7497 (moderate).
|
116 |
|
117 |
Here, more information on simplification operations will follow soon.
|
118 |
|
119 |
+
### Dataset Creation
|
120 |
|
121 |
+
#### Curation Rationale
|
122 |
|
123 |
DEplain-APA was created to improve the training and evaluation of German document and sentence simplification. The data is provided by the same data provided as for the APA-LHA data. In comparison to APA-LHA (automatic-aligned), the sentence pairs of DEplain-APA are all manually aligned. Further, DEplain-APA aligns the texts in language level B1 with the texts in A2, which result in mostly mild simplifications.
|
124 |
|
125 |
+
Further, DEplain-APA, contains parallel documents as well as parallel sentence pairs.
|
126 |
|
127 |
+
#### Source Data
|
128 |
|
129 |
+
##### Initial Data Collection and Normalization
|
130 |
|
131 |
+
The original news texts (in CEFR level B2) were manually simplified by professional translators, i.e. capito – CFS GmbH, and provided to us by the Austrian Press Agency.
|
132 |
All documents date back to 2019 to 2021.
|
133 |
Two German native speakers have manually aligned the sentence pairs by using the text simplification annotation tool TS-ANNO. The data was split into sentences using a German model of SpaCy.
|
134 |
|
135 |
+
##### Who are the source language producers?
|
136 |
+
The original news texts (in CEFR level B2) were manually simplified by professional translators, i.e. capito – CFS GmbH. No other demographic or compensation information is known.
|
137 |
|
138 |
+
#### Annotations
|
139 |
|
140 |
+
##### Annotation process
|
141 |
|
142 |
The instructions given to the annotators are available [here](https://github.com/rstodden/TS_annotation_tool/tree/master/annotation_schema).
|
143 |
|
144 |
+
##### Who are the annotators?
|
145 |
|
146 |
The annotators are two German native speakers, who are trained in linguistics. Both were at least compensated with the minimum wage of their country of residence.
|
147 |
They are not part of any target group of text simplification.
|
148 |
|
149 |
+
#### Personal and Sensitive Information
|
150 |
|
151 |
No sensitive data.
|
152 |
|
153 |
+
### Considerations for Using the Data
|
154 |
|
155 |
+
#### Social Impact of Dataset
|
156 |
|
157 |
Many people do not understand texts due to their complexity. With automatic text simplification methods, the texts can be simplified for them. Our new training data can benefit in training a TS model.
|
158 |
|
159 |
+
#### Discussion of Biases
|
160 |
|
161 |
No bias is known.
|
162 |
|
163 |
+
#### Other Known Limitations
|
164 |
|
165 |
The dataset is provided for research purposes only. Please check the dataset license for additional information.
|
166 |
|
167 |
+
### Additional Information
|
168 |
|
169 |
+
#### Dataset Curators
|
170 |
|
171 |
Researchers at the Heinrich-Heine-University Düsseldorf, Germany, developed DEplain-APA. This research is part of the PhD-program `Online Participation` supported by the North Rhine-Westphalian (German) funding scheme `Forschungskolleg`.
|
172 |
|
173 |
+
#### Licensing Information
|
174 |
+
|
175 |
+
The dataset (DEplain-APA) is provided for research purposes only. Please request access using the following form: [https://zenodo.org/record/7674560](https://zenodo.org/record/7674560).
|
176 |
|
177 |
+
#### Citation Information
|
178 |
|
179 |
+
If you use part of this work, please cite our paper:
|
180 |
|
|
|
181 |
|
182 |
+
```
|
183 |
+
@inproceedings{stodden-etal-2023-deplain,
|
184 |
+
title = "{DE}-plain: A German Parallel Corpus with Intralingual Translations into Plain Language for Sentence and Document Simplification",
|
185 |
+
author = "Stodden, Regina and
|
186 |
+
Momen, Omar and
|
187 |
+
Kallmeyer, Laura",
|
188 |
+
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics",
|
189 |
+
month = jul,
|
190 |
+
year = "2023",
|
191 |
+
address = "Toronto, Canada",
|
192 |
+
publisher = "Association for Computational Linguistics",
|
193 |
+
notes = "preprint: https://arxiv.org/abs/2305.18939",
|
194 |
+
}
|
195 |
+
```
|
196 |
This dataset card uses material written by [Juan Diego Rodriguez](https://github.com/juand-r) and [Yacine Jernite](https://github.com/yjernite).
|