Distance Learning and Inverse Mapping of Visualizations Applied to Text Mining

Abstract

With the increasing amount and complexity of data gathered and stored by computer systems, the task of exploring and extracting knowledge from it is becoming a hurdle. Aiming to solve this problem, data mining, machine learning, and information visualization fields have become of great importance. Recently, these fields were combined into a common new field, called Visual Analytics (VA), which comprehends the use of visual representations to control the creation of data mining and machine learning computational models, allowing users to incorporate knowledge into analytical tasks. Amongst the different types of data explored using VA techniques, text is a common and important type, as most of the human knowledge is stored in such format. This internship project proposes the application of the inverse mapping concepts studied in the original doctorate proposal in a text analytics environment, aiming at providing a better understanding of the effect of user-driven similarity metrics in VA text analytics systems.