Service: EU-JAV European Joint Action on Vaccination

The vaccine conversation monitoring platform was, in this initial phase of the project, developed in three languages: Italian, French and Spanish. The platform, accessible via credentials, monitors information flows from Twitter, Reddit, Wikipedia and Google Trends in real time. The platform allows the collection, analysis, aggregation and display of data extracted from different sources. This is possible thanks to the construction of the platform on three different levels (layers): the Data layer, the Analysis layer and the Visualization layer. The Data layer constantly collects raw data from the web and social media, pre-processes them through an initial screening and an initial cleaning of the raw data, obtained through the relative Application Programmer Interface, API: Twitter API, Reddit API, Google Trends API. To collect the data, keyword filters were used, created and validated through a structured framework, on the basis of recurrence and relevance criteria (6). The Analysis layer has the function of analyzing the processed and cleaned data through the generation of graphs, thus making the extracted data understandable and relating them to each other. The processes that allow the geolocalization of tweets take place on this layer, and two processes based on machine learning algorithms: Influencer analysis and Early-signal detection: Influencer analysis makes it possible to identify the most influential and most active users in the conversation about vaccines on Twitter, with the aim of monitoring the crucial "nodes" of the communities involved, both in vaccine-friendly communities and in the no vax community; Early-signal detection allows the interception of users who use new hashtags, therefore users new to the topic of vaccines. The hashtag (created by inserting the # key in front of the desired word or phrase, for example, #vaccini or #iononmivaccino) is a tool that serves as a thematic aggregator on some social networks and was introduced by Twitter as a keyword signal and to draw users' attention to some particular words or themes. Finally, the Visualization layer allows the visualization of the results, makes the extrapolated information understandable and communicates it in a simple but representative way, in order to understand the displayed data more easily.