Rethinking the data wheel: Automating open-access, public data on cyber conflict

摘要

To date, researchers studying cyber conflict through publicly available information sources have either selected on the actor or selected on the intrusion method when coding events. Both approaches lead to distinct challenges when it comes to result validation and the avoidance of selection bias. This article describes prospects for open-source, public data collection for cyber security events. We present an initial data collection and analysis effort of interstate cyber conflict incidents involving the United States as a pilot study. Using a tailored collection of more than 155,000 documents from print-only media sources, we describe a method to process data, parse document elements, and populate an event dataset. Human coders are then tasked with validation of incident information, after which the search code is updated to ensure greater accuracy in subsequent runs. In the study, the data produced are compared with previously available data on cyber conflict involving the United States. We demonstrate that the method can effectively capture and describe cyber conflict incidents for researchers to study in a broad range of research efforts. Moreover, this method captures greater granularity within cyber conflict episodes, which are inherently multi-faceted. This approach to cyber conflict analysis carries with it several distinct advantages over alternative research designs, in that it promises to produce significantly larger amounts of pertinent metadata than might otherwise be possible.