ThemeWeaver is an application that is built on top of the T2K module library. ThemeWeaver employs an intuitive user interface that guides users directly to important, isolated information typically discarded when a traditional statistical-based search engine is used. It delivers "clustered knowledge" to the user, enabling rapid information exploration and convergence on relevant information. ThemeWeaver is built on an object-oriented foundation based on semantic elements that are extracted from text and multimedia. Upon receipt of a query, ThemeWeaver will return a clustering of semantic elements from the data that are related to that initial query. The user can then interactively explore the meaningful elements in these clusters. The collection of queries and selected elements are called themes. ThemeWeaver will support users as they incrementally develop themes that are used to filter highly relevant "information nuggets" from within a large collection of documents. Once a theme has been identified, the system can automatically and asynchronously inform the user when new documents that fit the theme arrive. The themes can also be saved, so that they can re-query the data collection later.

We believe that this tool represents a new direction in query processing and reporting tools. In a world where data is rapidly accumulating, it will facilitate complex analysis of data in a multidimensional environment. In addition, ThemeWeaver will expand current knowledge discovery methodology by providing the means to discover critical information in multi-media data that was not known a priori. And finally, through the use of dynamic theme checking, ThemeWeaver will help to reduce the amount of time a user spends keeping up-to-date on a topic.

For more information on using ThemeWeaver™, contact:

Loretta Auvil, lauvil@ncsa.uiuc.edu