Vax network: profiling influential nodes with social network analysis on twitter

This paper analyses a social network based on clusters of people. The first cluster identifies people with opinion “pro-vax” the second one identified by “anti-vax” opinions emerged in the last three years. This analysis consists of a description of a data collection of public tweets available online; every collection is based on about one month of tweets per year. Tweets are extracted with specific “key-words” in the context of vaccine and anti-vaccine factors. We used twitter fetcher included in the semantic and social network analysis software; in this way we could fetch 120 days of tweets distributed over three years: apr-2015, nov-2016, jun-2017, corresponding on starting in the middle of 2015 of a relevant increasing of pro/anti-vaccine online information and the consequence information cycles at the end of 2016 and the middle of 2017. The total volume of collected tweets is about 300,000. All the collected tweets were written in English. In this elaboration we were able to distinguish behavior emerged in tweets in each cluster analysis by carrying out a preliminary sentiment analysis of the collected tweets. This network of Twitter, vax network, consists of about 800,000 links e relations. The analysis of vax network is based on the mainly centrality measures for identification of the relevant node in terms of potential influence of “vax networks”. Instead of the sentiment analysis is based on main methods of text mining of natural language used in the tweets of “vax network”, for profiling the relevant node in terms of its main characteristic correlated with own centrality in the network. Consequently we can identify the profiled nodes that have prevalent behavior for influencing or clustering the trend of the “vax network”.