Home // International Journal On Advances in Life Sciences, volume 5, numbers 1 and 2, 2013 // View article

Using Expert and Empirical Knowledge for Context-aware Recommendation of Visualization Components

Authors:
Martin Voigt
Martin Franke
Klaus Meißner

Keywords: visualization, recommendation, ontology, knowledge, collaborative filtering, mashup

Abstract:
Although many valuable visualizations have been developed to gain insights from large datasets, selecting an appropriate visualization for a specific dataset and goal remains challenging for non-experts. In this article, we propose a novel approach for knowledge-assisted, context-aware visualization recommendation. Therefore, both semantic web data and visualization components are annotated with formalized visualization knowledge from an ontology. We present a recommendation algorithm that leverages those annotations to provide visualization components that support the users' data and task. Since new visualization knowledge is generated while working with a visual analytics system due to users insights, particularly a component is suitable or not for a selected dataset, we track these findings by means of users explicit and implicit ratings. This empirical visualization knowledge is reused in subsequent recommendations to better adapt the ranking of components to users needs. We successfully proved the practicability of our approach by integrating it into a mashup-based research prototypes called VizBoard.

Pages: 27 to 41

Copyright: Copyright (c) to authors, 2013. Used with permission.

Publication date: June 30, 2013

Published in: journal

ISSN: 1942-2660