The article focuses on increasing the usefulness of extracted information. Extracting information from the audio track of news stories makes the data much more useful for the Informedia system. The Informedia Digital Video Library News on Demand system, initiated in 1994, which uniquely utilizes integrated speech and image and natural language understanding to process broadcast video, goes beyond the simple extraction of named entities from the audio stream and presents a demonstration that it is the ultimate use of this extracted information that matters, not the improved percentage point in extraction accuracy of one information extraction approach over another. It is emphasized that it is the integration and presentation of the extracted metadata that makes the Informedia video library usable and useful. It is critical to have multiple different types of errorfully extracted information, to allow the user to quickly scan of the information presented in different forms, so that metadata redundancy can overcome the errors of any one type of metadata