Design, development and usability evaluation of social system interface and development of computational model

Abstract

In recent times, methods of computational intelligence (CI) that aim to solve real-life problems are developed by computer science researchers in collaboration with domain experts. There has also been an increased emphasis on the usability aspect of these algorithms by developing easy-to-use web interfaces. The graphical user interfaces (GUIs) designed for these algorithms are often designed solely to connect the web interfaces to the algorithm’s functionality. While this is effective from researchers’ perspective, the needs of new users (such as policymakers) in relation to software use are often neglected. The lack of consideration of new users’ experience when developing GUIs often establishes usability issues for the technology and as a result expands the gap between the advances made in the computer science field and other fields, most notably the social sciences. This thesis investigates the various design, development, and evaluation methods for social simulation software and provides valuable insights for researchers and user interface designers who seek to create an effective GUI. Additionally, this thesis provides a case study of how computational models can be effectively applied for approaching complex social problems such as homelessness. In chapter 3 the development and testing process of the Homelessness Visualization (HOMVIZ) platform is discussed. The HOMVIZ platform uses a deep learning algorithm in order to predict potential trends in homeless populations in a particular area of interest. Various aspects of the user interface (UI) design were analyzed and a 14 participant usability testing session was conducted in order to discern the perceived usability of the platform. The UI evaluation session in this chapter involved software testing, focus groups, and questionnaires. These sessions provided our research with valuable qualitative and quantitative data. Chapter 4 explores moderated and unmoderated usability testing sessions and compares them in terms of efficiency, reliability, and flexibility. The research for this chapter was approved by the Lakehead University’s Research Ethics Board. The usability testing was conducted with a sample size of 72 participants. The research presented in this chapter provides valuable insight regarding different usability testing session methods and the impact of a known phenomenon called careless responding (CR) on data quality. Chapter 5 provides an example of how computational models can help mitigate a more complex social problem such as homelessness. The research presented in this chapter focuses on the operation of homeless shelters within Canada and introduces eight computation models that have the potential to improve the quality of life of people experiencing homelessness.