Abstract
Press releases covering novel scientific developments play an important role in science communication. Their number has increased in the past years, making them a valuable object of quantitative analysis in science communication research. However, large-scale research on academic press releases is limited, likely due to the lack of comprehensive and open datasets. In this data paper, we describe how we created a dataset of press releases published on EurekAlert!, an online science news service. The dataset consists of the metadata of a total of 566,566 press releases. We describe how we extracted the data, transformed it into a relational database, and made the code and data openly available to the research community. Next to that, we explain the structure of the dataset and how it can be accessed and propose potential integrations into other research information systems.