Headlines corpora with automatically extracted news values scores.

This dataset is part of a larger project on using headlines to predict the social media popularity of news articles. The first stage of the project was to develop methods of automatically extracting news values scores, which offer a new perspective on analysing headlines. The dataset consists of two headlines corpora -- The Guardian and New York Times -- collected in 2014 using news outlet APIs and used to develop the news values extraction methods. Each corpus includes a unique headline identifier (to enable recreating the corpus by querying the relevant API) and news values scores for each headline.