Overview: The Design, Adoption, and Analysis of a Visual Document Mining Tool for Investigative Journalists

Abstract

For an investigative journalist, a large collection of documents obtained from a Freedom of Information Act request or a leak is both a blessing and a curse: such material may contain multiple newsworthy stories, but it can be difficult and time consuming to find relevant documents. Standard text search is useful, but even if the search target is known it may not be possible to formulate an effective query. In addition, summarization is an important non-search task. We present Overview, an application for the systematic analysis of large document collections based on document clustering, visualization, and tagging. This work contributes to the small set of design studies which evaluate a visualization system "in the wild", and we report on six case studies where Overview was voluntarily used by self-initiated journalists to produce published stories. We find that the frequently-used language of "exploring" a document collection is both too vague and too narrow to capture how journalists actually used our application. Our iterative process, including multiple rounds of deployment and observations of real world usage, led to a much more specific characterization of tasks. We analyze and justify the visual encoding and interaction techniques used in Overview's design with respect to our final task abstractions, and propose generalizable lessons for visualization design methodology.

Speculative Practices: Utilizing InfoVis to Explore Untapped Literary Collections

“…This strategy has been extensively discussed in the visualization community to facilitate investigative analysis [45,53], the exploration of news articles [11], book collections [46], and artworks [21]. Studies have shown that the strategy of coordinated views is useful for the analysis of large and complex document collections, both in professional [5,18] and casual contexts [11,21]. Thudt et al highlight this strategy as particularly effective for supporting open-ended exploration and serendipitous discoveries [46].…”

“…Meanwhile in the visualization community long-term research processes and outcomes supported by visualization have been discussed in the area of journalism [5] and in casual scenarios [21]. What we lack are in-depth and long-term studies that show how information visualization can facilitate humanities research processes and how such visualizations integrate into traditional approaches such as archival work and close reading.…”

“…We introduce these as characteristics of a design space that applies to research on literary collections specifically, but also to humanities research at large. This space differs from related application contexts that have previously been discussed in the visualization research community, such as casual or artistic scenarios [33], libraries [46], museums [3,21], or journalism [5] by drawing from and expanding recent digital humanities discussions that seek to address critically the potential of information visualization in their research [13,16,17,25,27]. We posit four dimensions that characterize this space-the audience, research approaches within the humanities (quantitative analysis vs. interpretation), the role of visualization (result vs. process), and considerations on material and visual aspects of the source material.…”

“…Although digital humanities projects typically welcome public outreach, we know of no case study in the context of literary research where both public and academic audiences have been explicitly considered. Similarly, while case studies in information visualization have shown their suitability for casual, public audiences (e.g., [3,21,46]) on the one hand and for domain experts on the other (e.g., [5,18], to name a few), few address simultaneously a range of expert and casual audiences (see [18,29] for rare examples).…”

Visual exploration of topic variants through a hybrid biclustering approach

“…Il s'applique à des matrices binaires pour identifier des blocs dont les cellules sont uniquement composées de 1 (Figure 2b (Figure 4). En effet, l'exploration hiérarchique des sujets est plus efficace qu'une exploration plate basée sur des listes [16] et préférée par les journalistes utilisant Overview [7].…”