Professor Allan's research focuses on the automatic organization of textual information. He is interested in statistical methods extracting "concepts" from text and finding relationships between those concepts that can be used to impose structure that is otherwise hidden. For example, in the Topic Detection and Tracking work, the effort is toward adding topic-based structure to broadcast news sources as it evolves over time. He is also interested in information visualization, or methods for showing people the relationships between text or concepts--often when those relationships are not easily described by rules.