This chapter explores the architecture for a typical multimedia information mining system or methodology in the literature. It shows a specific multimedia information mining system and presents an example of a specific method on concept discovery through image annotation in a multimodal database. The chapter describes a probabilistic semantic model and the corresponding learning procedure to address the problem of automatic image annotation and show its application to multimodal image data mining and retrieval. It develops a probabilistic semantic model in which the visual features and textual words are connected via a hidden layer to constitute the concepts to be discovered to explicitly exploit the synergy between the two modalities. Multimedia information mining is an emerged, very interdisciplinary and multidisciplinary area involving three independent research fields: multimedia, information retrieval, and data mining. The information retrieval research concerns science and/or methodologies in searching documents or information within the documents.