In the last few years, terrorist organizations have been using social networks, specially
Twitter, with the purpose of creating fear, recruiting new members and indoctrinating.
This project has as its objectives the development of a classifier that allows us to de-
termine whether a tweet is radical or not using machine learning techniques, the implementation of a service for analyzing the radicalization of a text with our classifier and the execution of a dashboard for visualizing potential radical users.
For creating our classifier, we have developed a software program using the python
programming language and the tools that it provides for the processing of natural language (NLTK). For the feature extraction that we have considered (NER, POS, hashtags and sentiments) of the tweet of our database and for the use of machine learning algorithms for labeled data, we have used the library scikit-learn. As a result, we have obtained a model with an accuracy greater than 90% classifying the radicalization of the tweets contained in the database used in the project.ç
The implementation of the service for analyzing the radicalization of a text has been
carried out using the Senpy API. The development of a dashboard for visualizing possible radical users has been done using the environment Sefarad.