Datasets:

Modalities:
Text
Formats:
parquet
Languages:
Swedish
ArXiv:
License:
File size: 540 Bytes
5ba1f52
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
name: SwednClusteringP2P
description: The SWE-DN corpus is based on 1,963,576 news articles from the Swedish
  newspaper Dagens Nyheter (DN) during the years 2000--2020. The articles are filtered
  to resemble the CNN/DailyMail dataset both regarding textual structure. This dataset
  uses the category labels as clusters.
evaluation_framework: mteb
tasks:
- id: SwednClusteringP2P
- id: SwednClusteringP2P_default_summaries
  config: default
  split: summaries
- id: SwednClusteringP2P_default_articles
  config: default
  split: articles