Improving the use of open government data using visualizations

The steady increase in Open Government Data (OGD) initiatives has created both opportunities and challenges for a wide range of users including government employees, journalists, researchers, scientists and engineers. Currently more than one million datasets have been made available by governments around the world, at national, regional and local levels. These datasets cover all activities in which governments are involved, namely: political boundaries, transportation networks, education performance, health related data, budgets and financial reports. While these initiatives often report anecdotal success regarding improved efficiency and governmental savings, the potential applications of OGD remain a largely uncharted territory. In this work, we claim that there is an important portion of potential users who can benefit from the use of OGD, but who cannot do so because they cannot perform the essential operations needed to collect, process, merge, and make sense of the data. There are multiple reasons behind this, an important one being a fundamental lack of expertise and technical knowledge. To mitigate these problems we propose the use of visualizations as a medium to consume, share and interact with data. The use of diverse styles of visualization has proven useful for understanding large quantities of data in multiple fields, ranging from military to economics to basic science. The problem with existing visualization tools and techniques is that they treat visualizations as finished artifacts; except in rare situations, currently tools do not empower users to explore how data was used, from where it was obtained and how it was displayed. Current tools do not help users create derivative visualizations from existing ones, forcing users who want only a slightly modified version to create it from scratch. We claim that the use of visualizations can greatly enrich the use of open government data if these constraints are overcome. This thesis presents a study focused on facilitating the use of Open Government Data by people lacking a deep technical knowledge on how to use this data. One part of this study required identifying who these people are and discovering some of their most common problems related to the use of data, especially tasks that require the use of visualizations (whether to communicate the data, understand it, or any other function). It was also necessary for this thesis to create a tool that helped these people in different tasks that involve the use of visualizations and Open Government Data. The final part of this thesis presents the results of a user study showing how users can perform better in different tasks that involve the use of data using this tool. We will show that this tool provides a simpler environment for users to manipulate, create and share datasets and visualizations, leading to less effort and time on their part. The scope of this study includes stakeholders with an interest in Open Government Data including government employees, researchers and journalists. %At the moment of this writing, several people have been interviewed for the purpose of identifying their needs and the problems they have with the use of Open Government Data. The majority of these people are related or interested in OGD, whether as a producer or as a consumer. These users come from different backgrounds and have different levels of expertise and motivations to use OGD. Based on these interviews, I have defined a set of Personas (a well-known technique used in HCI and User Design) to identify the different actors in the Open Government Data Ecosystem. Several people related to Open Government Data were interviewed for the purpose of identifying their needs and the problems they have with the use of Open Government Data. These people are related or interested in OGD, whether as a producer or as a consumer. Based on these interviews, we defined a set of use cases and Personas (a well-known technique used in HCI and User Design) to characterize the different profiles that represent users in a Open Government Data Ecosystem. Based on these Personas and use cases, a web-based tool, OpenDataVis, was written that allows these users to create, explore and reuse visualizations based on Open Government Data. These visualizations expose not only the data; but they also allow users to show basic provenance metadata. This provenance metadata keeps accountability of how each visualization was created, where the data comes from, and when it was obtained. A user study was defined to compare the performance of people creating, analyzing and reusing visualizations using OpenDataVis versus their current favorite tool. Participants reported their experience of performing each task. We showed that in many cases users with basic technical knowledge can perform tasks in less time and with less effort using OpenDataVis than with conventional tools. We also showed than in other cases users could perform tasks that otherwise would not be possible for them to do with conventional tools. The main contributions of this thesis are: - Evidence of the interest from non-technical experts in the consumption, exploration and creation of visualizations based on Open Government Data. - A definition of an Open Government Data Ecosystem (OGDE) is and the different stakeholders involved in this ecosystem. - A characterization of stakeholders of an Open Government Data Ecosystem, their skills, knowledge and needs as data providers/consumers. - The creation of a tool that allow users to create, explore and reuse visualizations based on Open Government Data. - An evaluation of users' performance using this framework and comparing it with the current situation.