ABSTRACT
This chapter discusses the feasibility, techniques, and demonstrations of discovering hidden knowledge by applying multimedia duplicate mining methods to the massive amount of multimedia content. It outlines the visual and audio features are extracted from all the frames of a broadcast video stream. The chapter shows that some temporal recurrence fingerprints are proposed for capturing the temporal information of each recurring fragment pair. It explores three promising knowledge-discovery applications to show the benefits of duplicate mining. The first application is dedicated to fully unsupervised TV commercial mining for sociological analysis. The second application focuses on news story retrieval and threading: it uses a one-to-one symmetric algorithm with a local interest point index structure to accurately detect identical news events. The third application is for large-scale cross-domain video mining. A commercial film mining is an important and valuable task for competitive marketing analysis and is used as a barometer for advertising planning.
