NewsCube: A Multi-Dimensional HyperCube Representation for Interactive Exploration of Enterprise News Corpora

Abstract

In this paper, we demonstrate an intelligent, personalized user interface for enterprise users who are inundated with humongous amount of business relevant news data. These users need interfaces that enable fast and interactive exploration of large and evolving corpora of news articles. Existing user interfaces either focus on relative frequency or co-occurrences of keywords (e.g., word clouds, clustering, Topic modeling) or organize the news using timeline representation and its recent variants such as Story Forest. All these representations ignore that there are different latent aspects or dimensions about a news article which are of particular interest from an enterprise user's perspective. In this paper, we propose NewsCube, a multi-dimensional hypercube representation that overcomes the limitations of existing interfaces to explore news corpora. Comparison of NewsCube based user-interface with other UI representations by experienced business users and senior management has confirmed the ease of navigation, high utility and on-demand focused interactivity of the proposed NewsCube representation. The accompanying video gives an overview of the NewsCube.