Pervasive computing is giving rise to new opportunities for service organizations to gain insights into different social and economic activities and behaviors across large geographic areas (Sobolevsky et al., 2016; Zhou et al., 2017). Within this sphere, a number of major works have explored concepts such as urban analytics and human mobility analytics (Gonzalez et al., 2008; Barbosa-Filho et al., 2017; An et al., 2018; Xia et al., 2018). These bodies of research encompass a variety of unique perspectives and different focal outcomes. For example, Mohamed et al. (2017) used smart card data to understand urban mobility patterns in general, while D’Silva et al. (2018) used crowd-sourced mobile location data to predict demand for new physical businesses, and Zhou et al. (2017) used location-based social media data to examine the impact of government investment in cultural initiatives on economic development. What is not currently known however, is how pervasive computing technologies can be leveraged at the level of individual service firms as a basis for service analytics and ‘analytics as a service’.
The aim of this study is thus threefold. First, we review major developments in the pervasive computing domain and link these to major developments in service technologies. Second, we highlight some of the main areas in which pervasive computing platforms can be leveraged to provide insights for specific individual service firms. This spans what digital mobility patterns ‘say’ about different types of service consumers (in terms of their characteristics), as well as why different mobility patterns matter for different types of services. The latter is anchored in understanding the literal, ‘physical’ customer journey and the ‘metaphysical’ customer journey and service consumption experience. Third, we present results of a longitudinal field study, involving passive mobility tracking (with strictly opt-in participation) via a smartphone app in two European cities. We present preliminary results arising from this study based on unsupervised machine learning optimization techniques. Specifically, insights here are distilled through a combination of automated feature selection and subset selection based primarily on graph topology.
To support these goals, we also briefly introduce concepts from sensor fusion and algorithmic approximation as a basis for human activity recognition and mobility tracking. We also discuss important ethical and legal dimensions of human behaviour tracking at scale; a cornerstone of best-practice in digital service development, deployment and user acceptance in general. This focuses on GDPR and related data privacy requirements in the context of scaled technological infrastructure in general, including overlaps with considerations in the internet-of-things domain (Wachter, 2018). We conclude with a broad discussion of citizen-centric, data-enabled ecosystems, with an emphasis on using existing resources, including data, for improving quality of life (Delmastro et al., 2016; Ramaswami et al.,2016; Pereira et al., 2017; Yeh, 2017).