Understanding human-data interaction: Literature review and recommendations for design

Highlights

• An in-depth literature review of the Human-Data Interaction (HDI) area.

• Description of key research topics and open research challenges.

• Synthesis of HDI recommendations for design to meet good interactivity.

• Evaluation of HDI recommendations in Web portals with large amounts of data.

Abstract

The trend of collecting information about human activities to inform and influence actions and decisions poses a series of challenges to analyze this data deluge. The lack of ability to understand and interact with this amount of data prevents people and organizations from taking the best of this information. To investigate how people interact with data, a new area of study called “Human-Data Interaction” (HDI) is emerging. In this article, we conduct a thorough literature review to create the big picture about the subject. We carry out a variety of analyses and visual examinations to understand the characteristics of existing publications, detecting the most frequently addressed research topics and consolidating the research challenges. Based on the needs of HDI we found in the analyzed publications, we organize a set of recommendations and evaluate online systems that demand intensive human-data interaction. The obtained results indicate there are still many open questions for this interesting area, which is maturing with an increase number of publications in the last years, and that systems with large amount of data openly available poorly meet the proposed recommendations.

Introduction

In the last years, the world has seen a revolution in issues involving the tracking of information generated by the most diversified human activities. People's actions and preferences are being monitored in a myriad of ways. Collected data are used to influence decisions in almost all areas of individual and social life.

The understanding and analysis of large amounts of data play a critical role in everyone's empowerment. In this context, one of the greatest challenges is the complexity of the ecosystem within which data are produced, collected, edited, and used (Hornung et al., 2015). This varies from scenarios in which a single person is the producer, collector, editor, and user of data to scenarios demanding the involvement of many people. This complexity has given rise to questions such as: how to facilitate the understanding, manipulation, analysis, and sensemaking of large amounts of information?

Different fields of study are concerned with issues related to the interaction between people and data. Recently, the area “Human-Data Interaction” (HDI) has begun to investigate how people interact with data as an analogy with how Human-Computer Interaction (HCI) investigates the relationship between people and computers (Hornung et al., 2015).

Several relevant researches working on HDI have attempted to explain the objectives and frontiers of the area (Cafaro, 2012; Elmqvist, 2011; Hornung et al., 2015; Mortier et al., 2013, 2014). They introduced the problem, defined concepts, and presented some open challenges related to the subject. However, at the moment of writing this article, we are not aware of an updated and thorough literature review.

As a first contribution of this article, we provide an overview of the HDI research area based on a thorough literature review. We aggregate background information by analyzing publications and identifying key research topics. We use the guidelines proposed by Kitchenham et al. (2007) to support a systematic investigation into HDI. A systematic literature review is an organized approach to evaluate and interpret available research relevant to a particular research question, topic area, or phenomenon of interest by using a trustworthy, rigorous, and auditable methodology (Kitchenham et al., 2007). Our research has identified primary studies on the area by determining what, where, when, and who has published any material regarding the subject. We detected the most frequently approached aspects and classified the publications by well-defined criteria.

As a second contribution, we use the lessons learnt from the literature review to propose a set of HDI recommendations for information systems. These recommendations are based on the needs described by the studied publications. We organized them according to key aspects of interactivity with a pervasive computing environment including data representation, interaction with data, and data processing logic. We validated the applicability of the proposed set by evaluating online systems that demand intensive human-data interaction and have a large amount of data openly available. The evaluation focused on three websites containing data provided by the administration of the city of São Paulo to its citizens. According to the results, the proposed requirements are not fully satisfied.

The remaining of this study is organized as follows: in Section 2 we define the HDI concept; in Section 3 we detail the results obtained from the systematic literature review; In Section 4 we propose and evaluate a set of recommendations for HDI in information systems; in Section 5 we compile a series of open research challenges to be addressed in the area; in Section 6 we present a discussion about the obtained findings; and in Section 7 we summarize the conclusions.