Mobile Intelligence

21.7 SUMMARY

With the proliferation of mobile devices and multimedia data sources, it is in a great need of effective mobile multimedia services. However, with their unique constraints in display size, power supply, storage space as well as processing speed, the multimedia applications in mobile devices encounter great challenges. In this chapter, we present MoVR—a user adaptive video retrieval framework in the mobile wireless environment. While accommodating various constraints of the mobile devices, a set of advanced techniques are developed and deployed to address essential issues, such as the semantic gap between low-level video features and high-level concepts, the temporal characteristics of video events, individual user preference, and so on.

Specifically, HMMM scheme is proposed to model various levels of media objects, their temporal relationships, the semantic concepts, and high-level user perceptions. HMMM-based user profile is defined, which is also integrated seamlessly with a novel learning mechanism to enable the “personalized recommendation” for individual user by evaluating his/her personal histories and feedbacks. In addition, the fuzzy association concept is employed in the retrieval process such that the users gain the control of the preference selections to achieve reasonable tradeoff between the retrieval performance and processing speed. Furthermore, to improve the processing performance and enhance the portability of client-side applications, storage consumption ...