Towards Editorial Transparency in Computational Journalism

The increasing frequency of use of data and code in journalistic projects drives the need to develop guidelines or frameworks for how to responsibly and accountably employ algorithms and data in acts of journalism. One route to the accountable use of algorithms in journalistic work is to develop standards and expectations for transparency. In this paper we describe steps toward transparency with respect to computational journalism drawing from two case studies. The first case study concerns algorithmic accountability reporting where data collected via the Uber API and government sources were analyzed to understand quality of Uber service across Washington D.C. The second case centers on editorial transparency in the creation of a tool – in this case, a Twitter bot – built as an exploration of automated surfacing of anecdotal comments from news articles. Based on our experiences in these two cases we describe approaches to sharing data and code. The benefits of transparency as well as considerations such as licensing and documentation are discussed.