The lecture provides an overview of the latest developments in automatic summarization. It begins with an overview of summarization terminology and architecture concepts, and then examines human abstracting and its influence on automated summarization. The lecture next describes a variety of linguistic and statistical methods for producing extracts and abstracts, with an emphasis on empirical results. Multidocument and multilingual summarization are then discussed, followed by an overview of research on language models for summarization. The lecture concludes with a discussion of existing evaluation methods and results.

Dragomir R. Radev is Assistant Professor in the School of Information and the Department of Electrical Engineering and Computer Science at the University of Michigan. His research interests and publications are in text summarization, natural language generation, question answering, and Web-based information retrieval. In 2001, he led a Johns Hopkins Language Engineering summer workshop on Automated Text Summarization.