Automated personalization of multimedia presentations. We have created the abilty using implicit and explicit information from the user to automatically select the content and form of video presentation. This is embodied in the computational implementation, Personalized Broadcast News Nagivator. The essence of the invention is to exploit the fine-grained segmentation of broadcast video performed by BNN, instrumentation of user interactions, and the creation of a user model to then guide the content and form of the presentation of the news to the users interest and preferences. Since BNN automatically decomposes video into audio, video, and text streams (including summaries and extracts such as people, organizations, and locations), these can be recomposed in accordance with user preferences into a new artifact. The system both allows the user to explicitly state their preferences in the form of a media profile (e.g., audio, video, summaries, types of named entities) as well as a content profile (e.g. keywords or named entity specifications of the kinds of material they prefer.). It also monitors their interaction so that if they always choose a particular topic or a specific broadcast channel source, or data, or media type (e.g., summaries or images or both) the system will learn this from captured interaction logs. The system can then use this to present tailored content and media to the user.
Patent Number: 7,386,542
Date Issued: June 10 2008