Displaying Evolving Events Via Hierarchical Information Threads for Sensitivity Review

Many government documents contain sensitive (e.g. personal or confidential) information that must be protected before the documents can be released to the public. However, reviewing documents to identify sensitive information is a complex task, which often requires analysing multiple related documents that mention a particular context of sensitivity. For example, coherent information about evolving events, such as legal proceedings, is often dispersed across documents produced at different times. In this paper, we present a novel system for sensitivity review, which automatically identifies hierarchical information threads to capture diverse aspects of an event. In particular, our system aims to assist sensitivity reviewers in making accurate sensitivity judgements efficiently by presenting hierarchical information threads that provide coherent and chronological information about an event’s evolution. Through a user study, we demonstrate our system’s effectiveness in improving the sensitivity reviewers’ reviewing speed and accuracy compared to the traditional document-by-document review process.