Mapping the discourse on Smart Cities by means of NLP (Natural Language Processing) and complex network analysis

In the last decade or so, we witness two major revolutions that have a significant effect on our lives: the urban revolution and the big data revolution. Since 2008, more than half of the world's population lives in cities, while in the western world this percentage is considerably higher (above 80% in most of the Americas and Europe). At the same time, new technologies allow collecting data from bottom up customers and consumers and disseminating it back, using advanced technologies. The conjunction between this rapid urbanism and technological advances gave rise to a new domain in the academic and technological world known as "Smart Cities". Despite the growing interest in smart cities, a unique definition of the term is yet to be formed. The current work on smart cities is extensive and carried in many disciplines, while different disciplines focus on different aspects of this term. These range from studies that regard the smart city as a technology-oriented city (Batty, 2012) through others that emphasize green, sustainable properties of cities, and to works that stress the importance of human capital and high education Albino et al. (2015). Thus, the aim of this work is to map the discourses on Smart Cities in terms of topics, dynamics, and interdisciplinary trends. We propose to do so by bringing together (already used conjunction) two theoretical bodies: natural language processing (NLP) and complex networks theory. We identified the research fields that study smart cities as well as the top-ranked journals in these research fields. We extracted from these journals 641 papers from the fields of Technology, Urban Studies and Planning, Political Science, Law, Economics, and building and construction.
We analyzed these papers by means of Open CalaisTM - an NLP software that is general enough on the one hand (to identify topics from various fields) and on the other hand, accurate in extracting metadata from unstructured text. We used a specific metadata concept Open CalaisTM provides that is called "Social Tags". These tags are based on similarity of the content of the analyzed text to Wikipedia pages, and describe the topics discussed in the text. We used a co-word methodology on the "social tags" to create a network where the nodes represent topics that were identified as relevant to the content of each paper, and the links represent a relationship between two topics stemming from the fact that the two appeared in the same paper. To classify the different topics into categories that represent a higher level of the ontology, we used Wikipedia API to extract the categories that were linked to the examined titles. Our analysis focused on the network's dynamics, where we studied the change in the communities' structure and the topics that appeared in the K-cores over time.
Our findings suggest that the discourse on smart cities follows interdisciplinary trends and becomes more coherent within the different communities as well as between them. We have also found that while the discourse on smart cities focuses on the technological and economic aspects, issues that concern social justice or human rights are either absent or very marginal. This raises a significant question regarding the role of the academia in evaluating how innovations may affect society in general and how to prevent (or at least identify) the use of such innovations against the best interests of the human inhabitants of cities.
References