2018-09, Jentner, Wolfgang, Sacha, Dominik, Stoffel, Florian, Ellis, Geoffrey, Zhang, Leishi, Keim, Daniel A.

A fundamental task in criminal intelligence analysis is to analyze the similarity of crime cases, called comparative case analysis (CCA), to identify common crime patterns and to reason about unsolved crimes. Typically, the data are complex and high dimensional and the use of complex analytical processes would be appropriate. State-of-the-art CCA tools lack flexibility in interactive data exploration and fall short of computational transparency in terms of revealing alternative methods and results. In this paper, we report on the design of the Concept Explorer, a flexible, transparent and interactive CCA system. During this design process, we observed that most criminal analysts are not able to understand the underlying complex technical processes, which decrease the users’ trust in the results and hence a reluctance to use the tool. Our CCA solution implements a computational pipeline together with a visual platform that allows the analysts to interact with each stage of the analysis process and to validate the result. The proposed visual analytics workflow iteratively supports the interpretation of the results of clustering with the respective feature relations, the development of alternative models, as well as cluster verification. The visualizations offer an understandable and usable way for the analyst to provide feedback to the system and to observe the impact of their interactions. Expert feedback confirmed that our user-centered design decisions made this computational complexity less scary to criminal analysts.

2017, Sacha, Dominik, Jentner, Wolfgang, Zhang, Leishi, Stoffel, Florian, Ellis, Geoffrey

Criminal Intelligence Analysis (CIA) faces a challenging task in handling high-dimensional data that needs to be investigated with complex analytical processes. State-of-the-art crime analysis tools do not fully support interactive data exploration and fall short of computational transparency in terms of revealing alternative results. In this paper we report our ongoing research into providing the analysts with such a transparent and interactive system for exploring similarities between crime cases. The system implements a computational pipeline together with a visual platform that allows the analysts to interact with each stage of the analysis process and to validate the result. The proposed Visual Analytics (VA) workflow iteratively supports the interpretation of obtained clustering results, the development of alternative models, as well as cluster verification. The visualizations offer a usable way for the analyst to provide feedback to the system and to observe the impact of their interactions.

2016, Sacha, Dominik, Senaratne, Hansi, Kwon, Bum Chul, Ellis, Geoffrey, Keim, Daniel A.

Visual analytics supports humans in generating knowledge from large and often complex datasets. Evidence is collected, collated and cross-linked with our existing knowledge. In the process, a myriad of analytical and visualisation techniques are employed to generate a visual representation of the data. These often introduce their own uncertainties, in addition to the ones inherent in the data, and these propagated and compounded uncertainties can result in impaired decision making. The user's confidence or trust in the results depends on the extent of user's awareness of the underlying uncertainties generated on the system side. This paper unpacks the uncertainties that propagate through visual analytics systems, illustrates how human's perceptual and cognitive biases influence the user's awareness of such uncertainties, and how this affects the user's trust building. The knowledge generation model for visual analytics is used to provide a terminology and framework to discuss the consequences of these aspects in knowledge construction and though examples, machine uncertainty is compared to human trust measures with provenance. Furthermore, guidelines for the design of uncertainty-aware systems are presented that can aid the user in better decision making.

2015, Ellis, Geoffrey, Dix, Alan

Decision making under uncertainty can lead to irrational behaviour; such errors are often being referred to as cognitive biases. Related work in this area has tended to focus on the human’s analytic and sensemaking processes. This paper puts forward a novel perspective on this, proposing that some cognitive biases can also occur in the process of viewing visualisations. Consequently, this source of error may have a negative impact on decision making. This paper presents examples of situations where cognitive biases in visualisation can occur and outlines a future user study to investigate the anchoring and adjustment cognitive biases in visualisation.

2018, Ellis, Geoffrey

This book brings together the latest research in this new and exciting area of visualization, looking at classifying and modelling cognitive biases, together with user studies which reveal their undesirable impact on human judgement, and demonstrating how visual analytic techniques can provide effective support for mitigating key biases. A comprehensive coverage of this very relevant topic is provided though this collection of extended papers from the successful DECISIVe workshop at IEEE VIS, together with an introduction to cognitive biases and an invited chapter from a leading expert in intelligence analysis. Cognitive Biases in Visualizations will be of interest to a wide audience from those studying cognitive biases to visualization designers and practitioners. It offers a choice of research frameworks, help with the design of user studies, and proposals for the effective measurement of biases. The impact of human visualization literacy, competence and human cognition on cognitive biases are also examined, as well as the notion of system-induced biases. The well referenced chapters provide an excellent starting point for gaining an awareness of the detrimental effect that some cognitive biases can have on users’ decision-making. Human behavior is complex and we are only just starting to unravel the processes involved and investigate ways in which the computer can assist, however the final section supports the prospect that visual analytics, in particular, can counter some of the more common cognitive errors, which have been proven to be so costly.

2017, Sacha, Dominik, Jentner, Wolfgang, Zhang, Leishi, Stoffel, Florian, Ellis, Geoffrey, Keim, Daniel A.

VALCRI provides a challenging and overwhelming high-dimensional dataset that comprises of hundreds of extracted semantic features in addition to the usual spatiotemporal information or metadata. To overcome the curse of dimensionality and to generate low-dimensional representations of these semantic features we apply interactive high-dimensional data analysis techniques with the goal of obtaining clusters of similar crime reports. However, it is still a challenge for crime analysts to make sense of the results and to provide useful interactive feedback to the system. Therefore, we provide several tightly integrated interactive visualizations that allow the analysts to identify clusters of similar crimes from different perspectives and interactively focus their analysis on features or crime records of particular interest.

2015, Stoffel, Florian, Sacha, Dominik, Ellis, Geoffrey, Keim, Daniel A.

In this paper we describe a visionary system, VAPD, which supports crime analysts in uncertainty aware decision making in use of comparative case analysis. In this scenario, it is crucial for crime analysts to get an accurate estimate of uncertainties included in their data as well as those caused through data transformations and mappings, thus supporting analysts in calibrating their trust in the pieces of evidence gained through data analytics. VAPD consists of one data processing and three visualisation components that adopt a set of guidelines for handling uncertainties. The system focuses on conveying an accurate estimate of these uncertainties on processes and uncertainties that occur within its natural language processing components. Text data analysis is ambiguous and error prone, but is nevertheless an important part of the data analysis. Through its innovative handling of uncertainties, VAPD enables transparent and reliable decisions based on uncertainty-aware visual analytics.

2018, Ellis, Geoffrey

Despite more than 40 years of research into the field and the increasing use of the term in the media, there is still some uncertainty and even mystery over cognitive biases. This chapter provides a background to the topic with the aim to clarify what is meant by cognitive biases. After introducing some uses and misuses of the term, examples of common biases are presented. This is followed by a brief history of the research in the area over the years which illustrates the continued debate on cognitive biases and decision-making. Work in the emerging field of cognitive biases in visualization, prior to this publication, is outlined which concerns both the interpretation of the visualization and the visualization tools, such as visual analytic systems. Finally, we discuss the challenging issue of debiasing - how to mitigate the undesirable impact of cognitive biases on judgments.

2016, Jentner, Wolfgang, Ellis, Geoffrey, Stoffel, Florian, Sacha, Dominik, Keim, Daniel A.

The exploration of volumes of crime reports is a tedious task in crime intelligence analysis, given the largely unstructured nature of the crime descriptions. This paper describes a Visual Analytics approach for crime signature exploration that tightly integrates automated event sequence extraction and signature mining with interactive visualization. We describe the major components of our analysis pipeline — crime concept/event extraction, crime sequence mining, and interactive visualization. We illustrate its applicability with a real world use case. Finally, we discuss current problems, future plans, and open challenges in our development of a solution that incorporates automated event pattern mining with human expert feedback.

2015, Dix, Alan, Ellis, Geoffrey