Análise de comportamento do usuário em redes sociais veiculares

Abstract

Participation in social networks can provide a significant amount of data about users and their surroundings. When properly processed, such data can be used as an important source of information on human behavior if it provides reliable and quality information. In this work, we use a vehicular social network with the main objective of evaluating the impact of external factors on the users present in these environments through their contributions. We can see how speed of user and delay influence the reliability attributed to alerts. It also studies the tendency of improvement or degradation of the reliability attributed to the alerts of each user. It is possible to observe the association between pairs of alerts that occur on the same street in short intervals of time. We verify the interval of consecutive contributions of each user and the ratio of time interval between the first and last contribution and its total number of contributions. Results were obtained through a public dataset of the Waze application, available on the Internet. It was discovered that the most posted alerts are about congestion, and that users mostly do it during peak hours on weekdays and on weekends on the afternoon. It was found that users who move at higher speeds do not contribute to the network, and postings that present the longest delays to be published on the network are poorly evaluated. In addition, there was also a significant association between climate risk alerts and congestion. As the main result, it turned out that users who receive low reliability in their posts tend to keep score low on the following posts. Finally, it was also possible to notice that the interval between the contributions of each user has an average of 10 minutes and are not made daily to the social network, but when they do, the time interval between the alerts has a linear growth.