Data visualisation techniques for exploratory analysis processes

Abstract

The paper is concerned with the design of user-centered environments for the exploration of large datasets. The specific focus is on high frequency financial news and market data. In previous work, the ADAGE SOA based framework which allows users to model and execute analysis processes has been proposed. One disadvantage is that although much of a business process may be automated, there still remain many steps that are user driven, particularly at the visualization stage. This paper examines this problem in more detail and proposes the concept of a visualization instance (VI) that captures the four essential elements of most analysis processes: data sources, business processes, UI controls and display types. VIs offer an interactive way for the data to be dynamically generated while exploring the datasets. By providing the ability to compose several VIs, complex visualizations can be generated. The paper describes an implementation that allows the processing of Thomson Reuters news data based on an existing ADAGE platform. © 2013 Springer-Verlag.