While electronic communications have streamlined some knowledge work, they have also added to it by generating more to read. The DOCKET projects will develop a focused email information extraction and monitoring tool that will serve 4 functions: 1. Provide information management tools that extract and visualize entities such as people, places and things, and their relationships to each other. 2. Sorts email automatically by user-determined categories and visualize the results. 3. Visualize incoming email by categories generated on the fly for a quick overview of what has arrived. 4. Monitor subject, people and places to identify changes in patterns and alert the user when significant changes are identified. Docket will be natural language-based data mining software that will extract named entities such as people, places, and organizations automatically from documents. It will display a web of relationships between these entities graphically for quick viewing by the user with links to the original documents. Phase I will focus on user studies and a structural analysis of email text. It will develop a prototype for mining information from email messages.
