Authorship Profiling in Political Discourse on Twitter: Age and Gender Determination
Adam Skurla, Juraj Petrik · 2024
Author profiling is one of the important subtasks of stylometry. Profiling helps us learn as much as possible about the person on the other side of the communication just by analysing their text. Through this, we can determine the gender, age, occupation, or even religion of the person. In this paper, we focus on a method for determining the age and gender of the author. We use lexical, syntactic, semantic, and TFIDF features. In our method, we pay the most attention to data preprocessing, which is most important for stylometric tasks using classical machine learning methods. During preprocessing, we use several methods such as removing stopwords, lemmatization, deleting emojis. In this work, we also experiment with an imbalanced corpus and test various models. The goal is to have the best model possible for this specific task with good explainability.