An Interactive, Example-Based, Visual Clustering System

Abstract

This work describes and evaluates a novel interactive visual clustering system. It combines a 2D projection with a clustering algorithm that operates on this projected data. Users can interact directly through the 2D representation, by providing examples according to their expert ground truth. Each interaction incrementally updates the 2D projection and the associated clustering. Experiments show the effectiveness of the method, with as few as one interaction leading to a tangible influence on the visualization.

Interactive Clustering

“…In a sense, these papers consider the whole clustering process as a means to an end-clustering activity or clusters themselves are not an ultimate goal. Rather, they are useful for highlighting data of particular importance (e.g., Andrienko et al [6], Boudjeloud-Assala et al [19], Bruneau and Otjacques [22], Jiang and Canny [62], Lee et al [67], Rawlins et al [83], Seo and Shneiderman [88], Turkay et al [99]).…”

“…As users learn more about the data, they not only specify better parameters for any given task but also narrow down their goals. In Bruneau and Otjacques [22], the system provides the data projection and its clustering simultaneously in the same two dimensional space, so that a user may influence the clustering output by directly manipulating the spectral clustering projection (e.g., a user draws a line that separates red and blue clusters, then the system re-clusters based on that line from the projection view).…”

“…In Sourina and Liu [93], the tool provides various geometrical shapes (e.g., boxes and ellipses) to update the clustering parameters. Others include dragging and dropping data points from one cluster to another [21,33,47,66] and selecting data points that should/should not be in close proximity by interacting with the data points in a 2D projection [22,23,28,29,49].…”

“…users evaluate various constraints together with possible newer constraints [36]. Others argue that users should be part of the quality assurance process by incorporating user feedback into tool assessments (e.g., Bruneau and Otjacques [22]). Another open question is how to effectively integrate user feedback at varying abstraction and generalization levels, especially when feedback is coupled with cluster restructuring [54].…”

MCLEAN: Multilevel Clustering Exploration As Network

“…In doing this, consistency in the clustering results is improved. Similary, Bruneau & Otjacques (2013) proposed an approach to integrate user preferences into the clustering algorithm in an interactively way through 2D projection of the dataset. Rinzivillo et al (2008) proposed an exploratory methodology for exploring a large number of trajectories using clustering techniques.…”