MI-MERCURY: A mobile agent architecture for ubiquitous retrieval and delivery of multimedia information

• Authors: Nikolaos Papadakis;Anastasios Doulamis;Antonios Litke;Nikolaos Doulamis;Dimitrios Skoutas;Theodora Varvarigou
• Venue: Multimedia Tools and Applications
• Year: 2008

Mining multimedia information in the Web is in general an arduous task, due to the fact that, (a) humans perceive media content using high level concepts, (b) the subjective and vagueness of content interpretation, and (c) the fact that relevant data are often hidden in a huge amount of irrelevant information. In addition, delivering and distributing the retrieved information to a wide range of terminal devices of different properties over a wide range of networks to users of different preferences requires new tools and mechanisms for content transformation and adaptation. Other problems concern the language that the data are stored, which may not be the user's preferred language. To address these issues we propose an integrated, reconfigurable, adaptable and open architecture for mining, indexing and retrieving multimedia information based on a mobile agent technology scheme. The proposed architecture consists of three integral subsystems: the acquisition module, responsible for searching and retrieving media data (both textual and visual), the transformation module, able to adapt and transform the mined information to other forms of representation, and the distribution module for delivering and adapting the retrieved data in terms of terminal devices, network channels and user's preferences. The system is based on a reconfigurable architecture which is able to dynamically and automatically update the system response to user's actual needs and preferences, by extending descriptor classes that are considered more relevant by the users. New innovative algorithms are presented in this paper both at each system module as well as in the system integration. The system supports efficient content adaptation mechanisms, textual and visual summarization (both sequential and hierarchical), automatic language translation, ontological representation, visual processing and web-based data mining. Experimental analysis on real-life web sites has been performed to test the efficiency of the proposed scheme and compare it with other approaches presented in the literature.