Automatic Story Segmentation of Closed-Caption Text for Semantic Content Analysis of Broadcasted Sports Video

Sports videos can be characterized as a sequence of recurrent semantic story units. Storing sports videos in this story-unit-based form will lead to develop an intelligent content-based retrieval, browsing, and summarization system. The storage requires segmentation of videos and semantic understanding of each segment. Since transcribed broadcasted video speech, the closed-caption text, can be the useful information source for semantic indexing of each story unit, this paper proposes a method to automatically segment the closed-caption text of sports videos into the semantic units. The proposed method firstly tries to segment the speech transcript into the scene units, a set of which composes a story unit, in a probabilistic framework based on Bayesian networks. Finding the boundaries of the set of the scene units enables us to generate the story units in the close-caption. In this paper, we discuss some experimental results and the potentiality for utilizing them for indexing of the video and speech summarization.