Multi-facet information processing algorithms for news video based on event combing

As an important and effective media of communication, online news videos are being widely and incessantly consumed by innumerable audience through the Internet. However, identifying content of interest from massive online resources remains a quite challenging and time-consuming task due to the continuous explosive growth of news video data, increasingly diverse production modes and the shortage of powerful tools for efficient video navigation, which are seriously affecting user’s experience and information seeking efficiency. In this paper, a multi-facet information processing algorithm for news video based on event combing is reported. By establishing causal relationship and subordinate relationship between news events can help users comb news events and acquire the desired information from massive news video data both effectively and efficiently. The algorithms are mainly made up of three functional modules: the first one is the news timeline. The news videos is associated with their corresponding time nodes to reflect the times of occurrence of events. According to the sequence of news time nodes, users can better grasp the whole news events in the occurrence sequence of news events. The second is the news content keyword cloud module. It represents the content keywords of video within a specific time period, and helps users quickly appreciate the theme of video through the word cloud. The third is news topic clue clustering. According to the retrieval keywords entered by users, the system retrieves relevant videos from the database and carry out topic clustering operation to them. Every cluster is used to generate news clues in the form of card stack, reflecting the main content of the whole news event from different views. The advantages of the multi-facet information processing are demonstrated through benchmarked experimental results in comparison with peer methods.