Research Of Cross-media Retrieval Based On Probabilistic Method

With the development of computer and network technology, the amount ofinformation on the net has shown an explosive momentum, and the type of multimediavaries. Today, users are no longer satisfied with one modality retrieval, which leads to thedevelopment of semantic-based cross-media retrieval technology.The paper first clarifies the object of cross-media retrieval including multimodalmedia interaction retrieval and multimedia document retrieval, and discusses the two mainproblems in cross-media retrieval including content gab between different modalityobjects and semantic gab between low-level features and high-level emotion. To solvethese problems, dimension reduction, machine learning, et al. are available.Since the heterogeneous features can not be compared, the canonical correlationanalysis, which maps two variants into an isomorphic feature space with maximumcorrelation maintained, is applied to solve the content gab problem. To bridge the semanticgab, a classification model is employed to extract the semantic feature of multimediaobjects.When performing the multimodal media interaction retrieval, the retrieval algorithmnot only consider the semantic correlation between objects but also the similarity betweenthe objects of the same modality, which is implement by the construction of mediacorrelation graph and the introduction of transition probabilities between the objects of thesame modality. When performing the multimedia document retrieval, we fuse the semanticfeature of different modality objects into a single one, and the multimedia documentretrieval can be simplified as a single modality retrieval problem.Experiments show that the retrieval algorithm based on probability has certainadvantages in effectiveness and efficiency.