Our lead data scientist Andrea Perlato takes you on a journey through concepts like edge and node betweenness centrality, the chinese whispers algorithm, brokerage, closeness centrality and diffusion of innovation as he explores the concept of “smart cities”.
The concept of a Smart City holds much promise for improving the quality of life for city dwellers, but what exactly does “Smart City”mean?
During the last century, society has changed in its structural global level and the European Union is investing in research and innovation that are able to improve the quality of life of citizens making cities more sustainable.
The Smart City concept goes beyond the better resource use and less emissions. It means aligning information technologies (IT) to citizens’ needs. To understand needs of citizens, what could be more “smart” of exploring their opinions on Social Network? For this purpose, we developed a systematic study to shed more light about the opinion of citizen for making their cities more attractive.
Studying the Social Network structure in the field of Smart City, we were able to identify their importance and classifications. In fact, Smart City is an interdisciplinary concept that cover different aspects. For example, there is an extensive attention in what is called the “School of Future”, where the goal is to introduce students to the state-of-the-art to manage techniques for smart cities.
Cities are always the physical manifestation of the big forces at play: economic, environmental and social forces. Milan Expo 2015, the Universal Exposition scheduled from May 1 to October 3 in Italy, is a sort of small world tour. That is why we focused our analysis on Italian smart cities.
Additionally and more striking, we unveiled people that are able to influence trends. These people are often in “structural nodes” of the network, a sort of privileged paths that allows their ideas spreading rapidly.
It is evident here that studying the dynamics of the communication through Social Network could plays a key role in the development of the cities of the future, along with the others IT Service Management.
The end of cities, the rise of communities
Gallion and Eisner (1983)  defined the cities as a concentration of people in a geographic area that can support themselves in the economic activities. A city is not merely the built in time but also the product of human nature, where ideas can be transformed in a new energy (Gutkind 1962) .
A major global commercial fair – Milan Expo 2015 – has opened in northern Italy with food as the theme. Ten million tickets have been sold already for the six-month event.
Thanks to this attainment, Milan aims to become a “digital smart city”, the ideal place to convert ideas in new energy.
In this article, we want to explore how Expo 2015 has triggered the development of new projects in the smart city contest. Moreover, we will try to identify how the protagonists spread their ideas.
For this purpose, we decided to investigate Twitter Network, the major social community used in Italy as a privileged way to share observations in the field of smart city.
Crucially, we not intend here to explore the smart technology inside the fair – Expo 2015, but how Italian cities are modeling themselves thanks to this event.
Smart City Wheel Model
As a first assessment, we analyzed the tweets (based on Italy) which dealt the theme of smart city over the last year. To do that, we used the social media monitoring and analysis solution Crimson Hexagon.
The Cohen’s Smart Cities Wheel Model (Boyd Cohen, 2011)  is a holistic framework for considering all of the key components of what makes a city smart (Figure 1). We used it as a touchstone to classify Twitter’s comments.
Figure 1. The graph represent the Cohen’s Smart Cities Wheel, a model that offers six intelligence dimensions to consider in all smart cities: Smart economy, Smart government, Smart people, Smart living, Smart mobility, Smart environment. Text in white represent topics discussed on Twitter. Text in black represent topics not discussed on Twitter.
The graph above is a modified version of the Cohen’s Wheel, enriched with specific dimensions that we found on Twitter (in white text). In black text, we stressed the dimensions not sufficiently treated on the social media.
What emerged is the lack of attention of the social media on the measurement of resources. Our view is that Twitter should become the privileged place where to involve citizens in a dialog about waste of energy, highlighting periodically the total energy consumption (electricity per capita, CO2 emission) and the waste volume generated by the city per capita.
Open Data, are closely related with the concepts above. Therefore, we are not surprised that this argument is not treated. Finally yet importantly, a city cannot be defined “smart” if not take in to account the iniquity. Gini coefficient of inequity, developed by the Italian statistician and sociologist Corrado Gini, measures the inequality of income or wealth.
We believe that this technical data should become part of the education of citizens through social media. The market tends to produce cities that are not the reflection of consumer preferences incentive build vertically. If the municipal human resources gave attention to citizens’ opinion expressed on social media, the buildings developer would be bound by the approval process.
Although it is too early to tell what new forms of participation will exist, it seems desirable that smart and aware people on social media could be the key tool for spreading information favoring the participation of others.
On the other hand, innovation on green space, mobility and smart living are well discussed on Twitter.
One of the challenge regards the solution in climate change and environmental care. The school of energy of the University of Trento (Italy) organized the first edition of the IEEE Italy Section School on Future Energy Systems. The social Media gave a lot of attention on it, where the goal is to provide Ph.D. Students with state-of-the-art skills in this exiting and broad research field.
Another key priority expressed on Twitter was related to smart care. Bolzano (Italy) has developed a project called “Abitare Sicuri” (live safely) addressed to the elderly. The aim is to promote the philosophy of dignity and independence as it is about equipment and services.
The efficiency of such a system is not only based on technologies, but also integrating it with the involvement of all citizens. The great interest shown on the social media confirms that the experiment is on the right track.
Defining the stakeholders on Twitter
The second part of the article is dedicated to the identification of the stakeholders involved into creation of smart city. They could be individuals, organizations or collective group expressing any sort of interest on Twitter discussions. By integrating the prospective of stakeholders, it generates proactive changes on the vision of a smart city and new initiatives to follow.
We want to present the Social Network Analysis (SNA) on Twitter as a strategy for investigating social structures. The capability to find the stakeholders of the network could improve the possibility that the different partners archive an open dialogue and a common clear vision.
If we go back in the medieval times, certain cities became powerful and wealthy in part because they were located in good place, like near hub or near shipping ports.
Similarly, in Twitter network certain people (called node) are very powerful in part because they play a central role. Centrality tell us which people are important in the network. This goes beyond just looking at which people might be popular or looking at who is the best importer or exporter of comments, but crucially who are the people that bridge different region facilitating the dialogue.
First, we classified people using the Chinese Whispers Algorithm , the goal of this clustering approach is to partition the network into individual clusters that we colored, creating a more intuitive and easy interpreted visualization.
Now, we want to introduce the concept of brokerage . This is where a person lies other people creating a link between communities that otherwise might not be well connected. This kind of node is important because provide connectivity between not connected community. A generalization goes by name of Betweenness Centrality . In other words, Betweenness Centrality can be used to eliminate the digital divide with the aim of promoting an information society. This could became a sort of tool to promote sustainable development and facilitate resource promotion. The size of the nodes (people) of the graph below are proportional to the level of Node Betweenness Centrality, also the color refer the level of centrality of a connection (also called edge). Darken edges are privileged ways to reach many people – Edge Betweenness Centrality (Figure 2).
Figure 2. Simplify version of Social Network Graph of Smart City. The nodes represent Twitter @account or #hashtag. The size of nodes are proportional of the level of Node Betweenness Centrality, colors refer the cluster membership. Black edges have greater ability to link several clusters and the size is proportional to the degree of interactions between two nodes. The legend shows the main nodes for the ten largest clusters classified using the Chinese Whispers Algorithm.
Note: we normalized the centrality score using the size of the ego networks. The main reason for this is to allow the possibility to compare centrality scores across different groups.
The top 10 accounts are single actors (ego), and together with the nodes they are connected (alters), generate their own clusters. These networks are also known as the neighborhood networks or first order neighborhoods of ego. For simplicity, we will call it the network with groups.
Nodes labeled in the graph are instead accounts or hashtags able to glue different ego groups.
Burt (1992), in his book Structural Holes, provides ample evidence that having high Betweenness Centrality, which is highly correlated with having many groups, can bring benefits to ego. In our network, we can observe that the account @SmartCity4Italy has many nodes with high Betweenness. In particular, it uses many #hashtag that play a key role in the growth of its cluster, reaching the 34.1% of the entire network.
It is curious to note that the rest of top accounts are not aligned with the key terminology used by @SmartCity4Italy. For example, @SmartCity_Today prefers the usage of #smartcities instead of #SmartCity.
The extensive use of standard hashtags to express the same concept is of fundamental importance to decrease the distances between the groups. At the moment, the feeling is that the top accounts compete with each other, but the secret of experimental cities lies in the ability to spread their innovations.
There is in fact another level of resolution at which we could try studying: diffusion of innovation . The mechanisms of how new ideas diffuse through the lengths of the network. No matter here, whether a node is a brokering between different parts of the network, but we want to know which have easy access to a large part of the network to disseminate the information. Therefore, it is not about brokerage, but how far away is the rest of the network. By definition, the Closeness Centrality  is based on the length of the average shortest path between a node and all other nodes in the network. We have not limited our study to the calculation of this index of centrality, but the analysis here below explains which of the previous clusters contain users able to spread innovation discussed in the group to the rest of the network (Figure 3). In other word, we want to find out if there are clusters with significantly higher level of Closeness Centrality.
Figure 3. Bag Plot. It provides a bivariate version of the univariate boxplot. The bag contains 50% of all points. The points outside the fence in light blue are outliers (users with values of centrality extremely below the average).
We applied a multiple comparison procedure to answer the question about which clusters differ from the others. We found that the @veronasmartcity account is the best cluster (see supplemental material – Tukey HSD Test). As described on the graph above, its community contain many users with high level of Closeness Centrality (@carloreggiani, @cbenini, @michelebugliesi, @francescosacco, @CamillaOnTW, @alinomilan, @foodtechconect).
Thanks to the statistical Tukey HSD Test , we were able to identify the superiority in the dissemination of information by @veronasmartcity. The error in this statement this is less 5%.
Despite of the low percentage of size of the cluster (5.4%) Verona smart city community contain the higher degree of spreading information on the entire network.
The following (Figure 4) is a brief description of the most influential users that have the potential to make Verona the larger community within the smart city, surpassing giants as SmartCity4Italy and SmartCity_Today.
Figure 4. Best Evangelists obtained from the Verona Smart City community.
Of course, these people may also be recruited by other community as a social media evangelist. The job of evangelists is to spread information about a cause and keep people aware and informed. The approach described along the article, could serve as a privileged way to improve the productivity of companies along with others organizations. The ideas, bounce them off each other (engagement) and that results in action. If we make the flow of new ideas better, then the company or the organization works better.
In this article, we presented a very effective methodology to make cities more creative. We could take this approach to the entire population able to create inspired data-driven innovation. The key element to achieve this result is promoting communities with good communication. If they do not talk to each other or with the rest of the society, they risk an impoverishment.
The consideration described above, gave us a way to think about designing a network. Betweenness and Closeness Centrality provoke more engagement within communities and we can apply this metrics on any kind of networks.
We also introduced a statistical test (Tukey HSD Test) to figure out the differences between communities or clusters, with a confidence of 95%. The more significant communities can serve as a reference to promote the engagement.
It is important here to stress the fact that without this statistical test is not possible to understand if the clusters are different in terms of importance, leaving the interpretation to the intuition that is often incorrect.
Overall, these measures should be regarded as a crucial piece of evidence to shed light on the nature of the networked society and its underlying mechanisms.