%0 Journal Article
%T Utilizing Implicit User Feedback to Improve Interactive Video Retrieval
%A Stefanos Vrochidis
%A Ioannis Kompatsiaris
%A Ioannis Patras
%J Advances in Multimedia
%D 2011
%I Hindawi Publishing Corporation
%R 10.1155/2011/310762
%X This paper describes an approach to exploit the implicit user feedback gathered during interactive video retrieval tasks. We propose a framework, where the video is first indexed according to temporal, textual, and visual features and then implicit user feedback analysis is realized using a graph-based methodology. The generated graph encodes the semantic relations between video segments based on past user interaction and is subsequently used to generate recommendations. Moreover, we combine the visual features and implicit feedback information by training a support vector machine classifier with examples generated from the aforementioned graph in order to optimize the query by visual example search. The proposed framework is evaluated by conducting real-user experiments. The results demonstrate that significant improvement in terms of precision and recall is reported after the exploitation of implicit user feedback, while an improved ranking is presented in most of the evaluated queries by visual example. 1. Introduction In the recent years, the rapid development of digital technologies has led to the growing storage and processing capabilities of computers, as well as to the establishment of fast and advanced communication networks. Taking also into account the low cost of image and video capturing devices and the deep penetration of Internet in today¡¯s communities, large quantities of audiovisual content has become available and accessible worldwide. The availability of such content and the increasing user need of searching into multimedia collections place the demand for the development of advanced multimedia search engines; therefore, video retrieval remains one of the most challenging tasks of research. Despite the recent significant advances in this area, further advancements in several fields of video retrieval are required to improve the performance of current video search engines. More specifically, major research breakthroughs are still needed in the areas of semantic and interactive search possibly using multimodal analysis and retrieval algorithms, as well as relevance feedback [1]. The state-of-the-art video retrieval systems incorporate and combine several advanced techniques including text retrieval and visual content-based search, in order to support the user in locating video clips that meet their demands. One of the main challenges faced by these approaches is to generate efficient representations and descriptions of the video source. The initial step towards this direction is the video segmentation and indexing into smaller video
%U http://www.hindawi.com/journals/am/2011/310762/