Multimedia mining: towards the construction of a methodology and a non-structured date analytics tool

Abstract

This research addresses the development of multimedia mining projects by applying analytical techniques to texts, images, audio, and video. In order to develop these projects, a methodology to develop multimedia mining projects (Multimedia Analytical Methodology-MAM) is proposed. Likewise, the construction of a software tool (known as Multimedia Analytical Platform-PAM) which allows the analysis of multimedia mining is introduced. Methodology and platform are evaluated with two study cases on prediction of mammography abnormalities and analysis of medical imaging similarity. Results obtained allowed validating the steps proposed in the MAM methodology and using the PAM platform to extract the characteristics of medical images, to apply data mining techniques, and to satisfactorily evaluate the results obtained.