Abstract: The large size of video collection not only provides an easy way for users to share information, but also brings a big challenge for managing them, in particular online monitoring. A critical rectuirement to monitor the video information is to accurately and adaptively identify the key information describing the video, which is also the first step for video analysis and video search. In this paper, we focused on the extraction problem of the video information from different websites. Specifically,we proposed an engine framework for information extraction. We formally defined the description model in the framework and implemented a customizable engine for information. The proposed framework has been applied to a real-world application of a national department and obtains promising results. Experimental results show that the proposed approach can effectively extract the video information and it significantly outperforms the baseline methods.

Key words: General extraction engine framework, Internet video monitoring, Logical model of video website, Web information extraction, Algorithms for generating extraction patterns