Datasets:
Create README.md
Browse filesThe Combinations of Aligned data-sentenCes from nAturally PrOduced texts (hereafter: CACAPO) dataset is a dataset for data-to-text generation, which contains automatically scraped news reports for the sports, weather, stock, and incidents domain in English and Dutch, aligned with relevant attribute-value paired data, which were reverse-engineered from the articles. This dataset contains the original articles and attribute-value pairs for end-to-end training, where the training split has been manually altered to elongate 1504 (10% of the dataset) input records with contextually relevant attribute value pairings.
README.md
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
task_categories:
|
3 |
+
- text-generation
|
4 |
+
language:
|
5 |
+
- nl
|
6 |
+
- en
|
7 |
+
tags:
|
8 |
+
- E2E
|
9 |
+
- Dutch
|
10 |
+
- English
|
11 |
+
- Reverse Engineered
|
12 |
+
- RDF to Sentence
|
13 |
+
pretty_name: CACAPO_elongated_training
|
14 |
+
size_categories:
|
15 |
+
- 10K<n<100K
|
16 |
+
---
|