The need for summarization and personalization of summaries of media content has been driven by the recent and anticipated tremendous growth in the media world. We present our report on a panel which asked: who, why and when summarization is needed; what information should be summarized; and what forms should summaries take in order to understand the needs of summarization systems from the users' point of view.; To address the user needs for summarization systems, a generalized framework for summarizing structured programs is proposed. This framework can be adapted for different genres and different features that can be extracted. We illustrate this with summarization systems for different genres of structured programs and their applications.; The panel stressed that the summaries should be personalized. We propose a video summarization algorithm that uses a profile consisting of users' personality traits and the mapping to video features that different personality traits seem to prefer to generate a summary that is personalized for the user. We present a methodology and a supporting user study for generating user profiles that is obtained by mapping personality of users to content features that can be used to automatically create personalized summaries of broadcast television content. Three common personality profiles (Myers-Briggs, Merrill Reed, and Brain.exe) are elicited from 59 subjects, together with their preferred summary of news, music, and talk show videos. A factor analysis between the personality traits and the features in preferred summaries indicated that only some traits (e.g., gender, extraversion, control orientation, intuitiveness, etc.) and only some features (e.g., faces, reportage, text, chorus, host, etc.) had predictive value. The mapping of personality to feature is shown to differ by genre.; The personalized summaries are validated with user tests where subjects rated on a Likert scale of 1-5, two summaries side by side: one created for their personality profile (preferred summary) and one of the opposing personality profile (not-preferred summary). Thirty-two subjects give ratings for four videos each of news, talk shows, and music videos. The average for ratings for preferred summary for news, talk-shows, and music videos were 3.71, 3.32, and 3.16, respectively. The ratings for the not-preferred summaries were 3.24, 3.17, and 2.78, respectively. The analysis of ratings for each genre using ANOVA enables us to state that predominant difference for news videos and music videos come about because of difference in personalization.
展开▼
