Abstract
In this chapter, an interactive framework is developed to enable personalized news video recommendation and allow news seekers to access large-scale news videos more effectively. First, multiple information sources (audio, video and closed captions) are seamlessly integrated and synchronized to achieve more reliable news topic detection, and the inter-topic contextual relationships are extracted automatically for characterizing the interestingness of the news topics more effectively. Second, topic network (i.e., news topics and their inter-topic contextual relationships) and hyperbolic visualization are seamlessly integrated to achieve more effective navigation and exploration of large-scale news videos at the topic level, so that news seekers can have a good global overview of large-scale collections of news videos at the first glance. Through a hyperbolic approach for interactive topic network visualization and navigation, large amounts of news topics and their contextual relationships are visible on the display screen, and thus news seekers can obtain the news topics of interest interactively, build up their mental search models easily and make better search decisions by selecting the visible news topics directly. Our system can also capture the search intentions of news seekers implicitly and further recommend the most relevant news videos according to their importance and representativeness scores. Our experiments on large-scale news videos (10 TV news programs for more than 3 months) have provided very positive results.KeywordsAutomatic Speech RecognitionHyperbolic PlaneVideo ShotNews ProgramTopic NetworkThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.
Submitted Version (
Free)
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call