12. Virtual Tribes: Analyzing Attitudes Toward the LGBT Movement by Applying Machine Learning on Twitter Data

In this paper, we investigate the application of machine learning techniques in the context of social media. Specifically, we aim at drawing conclusions from users’ Twitter behavior and language to users’ attitudes toward the LGBT movement. By using an adjusted procedure of the Cross Industry Standard Process for Data Mining (CRISP-DM) process, we create a prediction model for investigating and identifying those attitudes. Furthermore, we formulate step-by-step instructions for its deployment. We provide the reader with a theoretical background for our research domain and describe the methods that we use. Results show that there are two groups of contrary attitudes toward the LGBT community and that the language and behavior of users in the groups, respectively, differ from each other. Also, we identify word analyses as a valuable means for prediction. We also apply our model on another dataset to investigate its interspersion with the previously identified groups and demonstrate its effectiveness for predicting attitudes of a single actor on Twitter. Finally, we critically assess our findings and propose further fields of investigation in this area.