Projects > Brussell


Brussell helps people track developments in ongoing news situations as they unfold over time. It uses general knowledge of situations such as kidnappings, court cases and mergers to retrieve and organize information about a particular situation of interest to the user.

Suppose you're reading an article about an event, such as a merger, product release or terrorist attack and you'd like to learn more about it. With Brussell, you can get an summary of the overall situation including information about the participants and events that took place. It uses models of situations called "scripts" to analyze the article you're reading, extract information and retrieve other articles to fill in the details of the situation summary. From the "bird's eye" view of the situation summary, you can inspect the participants and events to see the sentences and articles that describe them.

More generally, intelligence analysts, news junkies, and ordinary people all track developments in ongoing situations as they unfold over time and initiate queries to learn more about the past context of the events of interest to them. Brussell is an intelligent information system aimed at supporting this activity.

Brussell employs a combination of explicit semantic models, information retrieval (IR), and information extraction (IE) in order to track a situation. It finds relevant news stories, organizes those stories around the aspects of the situation to which they pertain, and extracts certain basic facts about the situation for explicit representation.

Project Papers

  • Rich Interfaces for Reading News on the Web
  • Using Explicit Semantic Models to Track Situations across News Articles