Markovian image retrieval

This research proposes a new methodology for the retrieval of images which is based on Markov chains and its ability to store the semantics of the present situation in its states. The new algorithm called Koogle, creates a Global Markov Chain for keyword relevance and for storing the user semantics, where each state can hold more than one keyword. Since this model takes into consideration the targeted user preferences, it proves to be a better approach than most of the present methods. The research shows how multiple keywords can be included in a single state and how the new proposed ranking algorithm produces relevant results in most cases and can be used to give appropriate results for its targeted users.