Event data offers rich insights into political phenomena, but its use is hindered by the labor-intensive creation of dictionaries for coding events. This paper introduces an automated approach that generates dictionary entries using minimal human input—a small sample dictionary. By leveraging advances in natural language processing and machine learning, it significantly reduces researcher effort needed to define new domains for event coding. The method enables researchers to quickly translate domain interests into structured event datasets without needing comprehensive prior dictionaries. To demonstrate its effectiveness, we created a novel cybersecurity incident dataset using this approach.
