Pashtu Text Sentiment Analysis Dataset

Published: 4 March 2024| Version 1 | DOI: 10.17632/s9k7dk9sc6.1
Aizaz Ali


We gathered information from respected Pashtu newspapers, such as those published by the Wahdat Newspaper, and articles written by professors from various areas. This data collection took place over a month, encompassing regular newspaper issues and contributions from professors. All text was acquired digitally from faculty members of the Pashtu department and the Wahdat Newspaper. Initially, we amassed a total of 29,000 sentences in their raw form. Later, we conducted further processing, including cleaning and preprocessing, resulting in around 21,800 sentences of refined data. Subsequently, domain experts meticulously examined the data and attributed sentiments to each sentence.



Pak-Austria Fachhochschule Institute of Applied Sciences and Technology


Natural Language Processing, Machine Learning, Information-Processing of Emotion, Sentiment Analysis