| Resumen |
Bots identification has gained relevance within social networks due to its ability to influence the opinion of users on political, consumer and ideological issues. This is why research related to bot identification has grown in recent years. Various models have been proposed for the identification of bots, but this is an issue that has not been resolved yet. In this article, a model is proposed that, through the use of specific preprocessing and a four-layer neural network, improves the bot-human classification accuracy of Twitter messages, reaching a precision of 0.9462, which represents an advance with respect to what is presented in the state of the art with the same corpus. © 2020, Springer Nature Switzerland AG. |
