Romanian Categorized Web Dataset

Around 5.000 categorized and sentiment annotated Romanian web articles.

Gathering the data

Due to copyright reasons, the data is not included in the corpus. You will need to run a Python script in order to download and parse the text data from the URLs included in the ro_index.csv file.

The script is ran as follows:

pip install -r requirements.txt
python gather.py

The text data will be saved in the ./data folder. The script sends the URL to a web service that downloads the web page contents and properly strips away the HTML tags, leaving only the cleaned text.

Categories

• Arta si Traditii
• Astrologie
• Automobile
• Business
• Divertisment
• DIY
• Economic
• Educatie
• Familie
• Gastronomie
• Istorie
• IT&C
• Legal
• Media
• Medical
• Meteo
• Moda & Frumusete
• Monden
• Natura & Ecologie
• Politic
• Relatii si Sex
• Religie
• Sanatate
• Social
• Sport
• Stiinta & Tehnologie
• Turism & Calatorii

Sentiment Labels:

• Negative
• Neutral
• Positive