Network analysis in journalism: Practices and possibilities

Author:
Jonathan Stray, School of Journalism, Columbia University

Abstract:
Network analysis has been proposed and taught as an enabling technology for computational journalism. We report on a set of 34 stories involving network analysis, including three in-depth case studies and a breakdown of how networks were obtained, analyzed, and presented. Networks were commonly extracted from unstructured documents like public records. Visualization is the main analysis method and a popular way to present results. Algorithmic techniques such as centrality methods and community detection are known to journalists but seldom used, in part because the results are hard to interpret and contextualize. On the cutting edge of investigative journalism, graph databases are emerging as a powerful and flexible technology suitable for fusion of diverse data sets. However, these databases cannot be directly visualized because they are dirty and overwhelmingly large. Based on journalist interviews and the experience of previous stories, we explore the long practice of making network "sketches" or "maps." These are not so much data visualizations but conceptual tools that investigators use to think through a story. Maintaining a mapping between these sketches and the underlying graph database is a major design challenge. We propose a system that integrates a deep graph data store with interactive record linkage and network sketching.

More on http://www.kdd.org/kdd2017/

KDD2017 Conference is published on http://videolectures.net/

Видео Network analysis in journalism: Practices and possibilities канала KDD2017 video