dutch_social / README.md
albertvillanova's picture
Replace YAML keys from int to str (#2)
  - machine-generated
  - crowdsourced
  - en
  - nl
  - cc-by-nc-4.0
  - multilingual
  - 100K<n<1M
  - original
  - text-classification
  - sentiment-classification
  - multi-label-classification
pretty_name: Dutch Social Media Collection
    - name: full_text
      dtype: string
    - name: text_translation
      dtype: string
    - name: screen_name
      dtype: string
    - name: description
      dtype: string
    - name: desc_translation
      dtype: string
    - name: location
      dtype: string
    - name: weekofyear
      dtype: int64
    - name: weekday
      dtype: int64
    - name: month
      dtype: int64
    - name: year
      dtype: int64
    - name: day
      dtype: int64
    - name: point_info
      dtype: string
    - name: point
      dtype: string
    - name: latitude
      dtype: float64
    - name: longitude
      dtype: float64
    - name: altitude
      dtype: float64
    - name: province
      dtype: string
    - name: hisco_standard
      dtype: string
    - name: hisco_code
      dtype: string
    - name: industry
      dtype: bool_
    - name: sentiment_pattern
      dtype: float64
    - name: subjective_pattern
      dtype: float64
    - name: label
            '0': neg
            '1': neu
            '2': pos
  config_name: dutch_social
    - name: train
      num_bytes: 105569586
      num_examples: 162805
    - name: test
      num_bytes: 35185351
      num_examples: 54268
    - name: validation
      num_bytes: 34334756
      num_examples: 54269
  download_size: 68740666
  dataset_size: 175089693

Dataset Card for Dutch Social Media Collection

Table of Contents

Dataset Description

Dataset Summary

The dataset contains 10 files with around 271,342 tweets. The tweets are filtered via the official Twitter API to contain tweets in Dutch language or by users who have specified their location information within Netherlands geographical boundaries. Using natural language processing we have classified the tweets for their HISCO codes. If the user has provided their location within Dutch boundaries, we have also classified them to their respective provinces The objective of this dataset is to make research data available publicly in a FAIR (Findable, Accessible, Interoperable, Reusable) way. Twitter's Terms of Service Licensed under Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) (2020-10-27)

Supported Tasks and Leaderboards

sentiment analysis, multi-label classification, entity-extraction


The text is primarily in Dutch with some tweets in English and other languages. The BCP 47 code is nl and en

Dataset Structure

Data Instances

An example of the data field will be:

  "full_text": "@pflegearzt @Friedelkorn @LAguja44 Pardon, wollte eigentlich das zitieren: \nhttps://t.co/ejO7bIMyj8\nMeine mentions sind inzw komplett undurchschaubar weil da Leute ihren supporterclub zwecks Likes zusammengerufen haben.",
  "text_translation": "@pflegearzt @Friedelkorn @ LAguja44 Pardon wollte zitieren eigentlich das:\nhttps://t.co/ejO7bIMyj8\nMeine mentions inzw sind komplett undurchschaubar weil da Leute ihren supporter club Zwecks Likes zusammengerufen haben.",
  "created_at": 1583756789000,
  "screen_name": "TheoRettich",
  "description": "I ❤️science, therefore a Commie.   ☭ FALGSC: Part of a conspiracy which wants to achieve world domination. Tankie-Cornucopian. Ecology is a myth",
  "desc_translation": "I ❤️science, Therefore a Commie. ☭ FALGSC: Part of a conspiracy How many followers wants to Achieve World Domination. Tankie-Cornucopian. Ecology is a myth",
  "weekofyear": 11,
  "weekday": 0,
  "day": 9,
  "month": 3,
  "year": 2020,
  "location": "Netherlands",
  "point_info": "Nederland",
  "point": "(52.5001698, 5.7480821, 0.0)",
  "latitude": 52.5001698,
  "longitude": 5.7480821,
  "altitude": 0,
  "province": "Flevoland",
  "hisco_standard": null,
  "hisco_code": null,
  "industry": false,
  "sentiment_pattern": 0,
  "subjective_pattern": 0

Data Fields

Column Name Description
full_text Original text in the tweet
text_translation English translation of the full text
created_at Date of tweet creation
screen_name username of the tweet author
description description as provided in the users bio
desc_translation English translation of user's bio/ description
location Location information as provided in the user's bio
weekofyear week of the year
weekday Day of the week information; Monday=0....Sunday = 6
month Month of tweet creation
year year of tweet creation
day day of tweet creation
point_info point information from location columnd
point tuple giving lat, lon & altitude information
latitude geo-referencing information derived from location data
longitude geo-referencing information derived from location data
altitude geo-referencing information derived from location data
province Province given location data of user
hisco_standard HISCO standard key word; if available in tweet
hisco_code HISCO standard code as derived from hisco_standard
industry Whether the tweet talks about industry (True/False)
sentiment_score Sentiment score -1.0 to 1.0
subjectivity_score Subjectivity scores 0 to 1

Missing values are replaced with empty strings or -1 (-100 for missing sentiment_score).

Data Splits

Data has been split into Train: 60%, Validation: 20% and Test: 20%

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

The tweets were hydrated using Twitter's API and then filtered for those which were in Dutch language and/or for users who had mentioned that they were from within Netherlands geographical borders.

Who are the source language producers?

The language producers are twitter users who have identified their location within the geographical boundaries of Netherland. Or those who have tweeted in the dutch language!


Using Natural language processing, we have classified the tweets on industry and for HSN HISCO codes. Depending on the user's location, their provincial information is also added. Please check the file/column for detailed information.

The tweets are also classified on the sentiment & subjectivity scores. Sentiment scores are between -1 to +1 Subjectivity scores are between 0 to 1

Annotation process

[More Information Needed]

Who are the annotators?

[More Information Needed]

Personal and Sensitive Information

As of writing this data card no anonymization has been carried out on the tweets or user data. As such, if the twitter user has shared any personal & sensitive information, then it may be available in this dataset.

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

Dataset provided for research purposes only. Please check dataset license for additional information.

Additional Information

Dataset Curators

Aakash Gupta Th!nkEvolve Consulting and Researcher at CoronaWhy

Licensing Information

CC BY-NC 4.0

Citation Information

@data{FK2/MTPTL7_2020, author = {Gupta, Aakash}, publisher = {COVID-19 Data Hub}, title = {{Dutch social media collection}}, year = {2020}, version = {DRAFT VERSION}, doi = {10.5072/FK2/MTPTL7}, url = {https://doi.org/10.5072/FK2/MTPTL7} }


Thanks to @skyprince999 for adding this dataset.