Decoupling the Information Application from the Information Creation: Video as Learning Objects in Three-Tier Architecture

In this paper a new three-tier video application infrastructure is proposed. Video learning object is introduced as the intermedia tier and it connects the video metadata and video application tier. In addition to the traditional text metadata such as title and description, a wide range of visual metadata including key frames, video storyboards, and fast-forward video are integrated into the video learning object. The visual metadata is organized in hierarchical structure. Video timestamps are used to “mark” different video channels (i.e., visual channel, audio channel, text channel) for the purpose of synchronization. A MPEG-7 compatible XML metadata schema was developed to encode the video learning objects to achieve the reusability and interoperability—different video applications can use the same video learning object without having to reinvent the wheel. A semi-automatic video metadata authorization system called Video Annotation and Summary Tool (VAST) was developed to facilitate the creation of video metadata. The automatic process generates a series of key frame “candidates” (primitive frames) but human beings have to manually select the key frames. The number and the image size of the primitive frames are reconfigureable. The fast-forward version of the original video can also be created from the primitive frames. Finally, we demonstrate how the new three-tier infrastructure is used to guide the development of two independent video application projects: the Open-video digital library project and distance learning using the Interactive Shared Educational Environment (ISEE) project. Both projects utilize the same set of video learning objects.