Economic Relevant News from The Guardian

Published: 27 December 2019| Version 3 | DOI: 10.17632/yt8j2f3hpp.3


The news: The present dataset consists of 1789 news articles from the British daily newspaper The Guardian extracted using the content endpoint of The Guardian Open Platform. The news articles were, at the time, all the news corresponding to the sections: business, politics, society and world news for the entire month of January of 2013 (for a total of 1689 news) and an extra set of news articles randomly selected from the period Febrary of 2013 to December of 2015 (100 news articles). The first set of 1689 news articles was used for training and the second set of 100 news articles was used for testing in two publications: * Maisonnave, M., Delbianco, F., Tohmé, F.A. and Maguitman, A.G., 2018, November. A Supervised Term-Weighting Method and its Application to Variable Extraction from Digital Media. In XIX Simposio Argentino de Inteligencia Artificial (ASAI)-JAIIO 47 (CABA, 2018). * Maisonnave, M., Delbianco, F., Tohmé, F.A. and Maguitman, A.G., 2019. A Flexible Supervised Term-Weighting Technique and its Application to Variable Extraction and Information Retrieval. Inteligencia Artificial, 22(63), pp.61-80. The labels: The entire dataset was manually classified into two possible categories: economically relevant and irrelevant. The labelling process was carried out by two experts in Economy working in collaboration. For each news article, the full text of the article was analyzed to determine the category. The format: There are two different versions for this dataset: the reduced and the full versions. The former consists of a CSV and a readme file. The CSV file has five columns: "Instance No.", "Title", "Web Publication Date", "web URL" and "Economically Relevant". This version is reduced in columns as it does not include the full article texts; however, it does include all the 1789 instances. Requesting the full dataset: To gain access to the full version of the dataset (which includes the body of the news articles), please send an email to with a copy to requesting authorization and making it clear that the data set will not be used for commercial purposes.



Universidad Nacional del Sur


Natural Language Processing