There is an inability of the traditional public administration to effectively listen and understand citizens’ needs, together with a lack of a more qualitative and effective monitoring of public services. This has led to a search for new methodologies of knowledge management, discovering and monitoring how public decision-making is perceived from the city and citizens. Most of the literature on knowledge management has focused on the creation, acquisition, transfer and value creation within a public administration, but comparatively little work has been done to understand the management of knowledge of social media.
Following the previous challenge, the topic of sentiment analysis refers to a knowledge discovery process based on techniques for collecting, processing and analysing textual data to detect assessments, opinions and, more generally, subjective expressions. Furthermore, the progressive spread of social networks (such as Facebook and Twitter) has made available a huge amount of data on the preferences and opinions of citizens who use social platforms to search and share information and experiences.
Reasons for the adoption of sentiment analysis tools are to be searched in the image (sentiment) perceived by citizens in relation to a specific theme. The big amount of data generated by social media is converted into knowledge and thus provides an unparalleled strategy able to orient and guide choices and actions.
Through the use of mathematical and statistical methods, unstructured data are used to analyse information with the aim of quantifying the intensity (positive/negative) of a sentiment described in natural language in a text. The added value of sentiment analysis, compared to the usual customer satisfaction techniques, lies in the fact that it is about listening to emotions that are spontaneously decorated, thus naturally reflecting the real expectations and moods of citizens.
According to this approach, a critical component to be solved for a change in the modern public administration is the ability to take advantage of all information available into the territory. There are constantly arising sources of social big data, where public decision makers are called to analyse in almost real time.
The public sector’s knowledge about citizens’ needs and high priority social challenges, as well as the access to this information for statistical and analysis purpose, are of great value for the Municipality of Bari. These can be crucial in determining the opportunity of developing and addressing new investments and strategies in a given urban context. The effort to search for relations between external social data and internal information should be incremental and continuous, to produce new added-value knowledge.
Thus, the objective of the project is the realisation of an artificial intelligence solution, the Sentiment and Monitoring Analysis Tool for Cities and Citizens (SMACC), to improve public decision-making according to what people are talking about on social channels. It will implement a smartness model through the adoption of a multidimensional monitoring systems, built on a large set of performance indicators for the analysis and the processing of add-value scenarios of the urban context.
The data collection is coming from the web, from specific social channels related to the city of Bari and the lifestyle of its citizens. This is integrated within the Urban Control Centre in a common data lake, where structured and unstructured meet each other to accomplish a unique goal. This process generates and makes available new information able to support to public decision makers in the decision-making process of the urban governance. It makes them aware of topics of discussion in the web and how their actions impact on the community.
One of the first results of the SMACC solution was a case study on the concept of urban security. In this scenario, the extraction of the quantitative data and the subsequent activities aimed at the qualitative assessment of the perception of topics of discussion were complemented by analyses carried out with interpretative models. The models were developed with the support of sociologists and statisticians that, due to peculiarities and complexity of the data, supported this phase of the project to improve the quality of the investigations. In this way, an additional element of value was provided for the subsequent decision-making phases.