%0 Journal Article %T Video Pulses: User-Based Modeling of Interesting Video Segments %A Markos Avlonitis %A Konstantinos Chorianopoulos %J Advances in Multimedia %D 2014 %I Hindawi Publishing Corporation %R 10.1155/2014/712589 %X We present a user-based method that detects regions of interest within a video in order to provide video skims and video summaries. Previous research in video retrieval has focused on content-based techniques, such as pattern recognition algorithms that attempt to understand the low-level features of a video. We are proposing a pulse modeling method, which makes sense of a web video by analyzing users' Replay interactions with the video player. In particular, we have modeled the user information seeking behavior as a time series and the semantic regions as a discrete pulse of fixed width. Then, we have calculated the correlation coefficient between the dynamically detected pulses at the local maximums of the user activity signal and the pulse of reference. We have found that users' Replay activity significantly matches the important segments in information-rich and visually complex videos, such as lecture, how-to, and documentary. The proposed signal processing of user activity is complementary to previous work in content-based video retrieval and provides an additional user-based dimension for modeling the semantics of a social video on the web. 1. Introduction The web has become a very popular medium for sharing and watching video content [1]. Moreover, many organizations and academic institutions are making lecture videos and seminars available online. Previous work on video retrieval has investigated the content of the video and has contributed a standard set of procedures, tools, and data-sets for comparing the performance of video retrieval algorithms (e.g., TRECVID), but they have not considered the interactive behavior of the users as an integral part of the video retrieval process. In addition to watching and browsing video content on the web, people also perform other ˇ°social metadataˇ± tasks, such as sharing, commenting videos, replying to other videos, or just expressing their preference/rating. User-based research has explored the association between commenting and microblogs, primarily tweets, or other text-based and explicitly user-generated content. Although there are various established information retrieval methods that collect and manipulate text, they could be considered burdensome for the users, in the context of video watching. In many cases, there is a lack of comment density when compared to the number of viewers of a video. There are a few research efforts to understand user-based video retrieval without the use of social metadata. In our research, we have developed a method that utilizes more so implicit user interactions for %U http://www.hindawi.com/journals/am/2014/712589/