Interactive Metric Learning-Based Visual Data Exploration: Application to the Visualization of a Scientific Social Network

Data visualization is a core approach for understanding data specifics and extracting useful information in a simple and intuitive way. Visual data mining proceeds by projecting multidimensional data onto two-dimensional (2D) or three-dimensional (3D) data, e.g., through mathematical optimization and topology preserved in multidimensional scaling (MDS). However, this projection does not necessarily comply with the user’s needs, prior knowledge and/or expectations. This paper proposes an interactive visual mining approach, centered on the user’s needs and allowing the modification of data visualization by leveraging approaches from metric learning. The paper exemplifies the proposed system, referred to as Interactive Metric Learning-based Visual Data Exploration (IMViDE), applied to scientific social network browsing.