A Word Embeddings based Approach for Author Profiling: Gender and Age Prediction

Author Profiling (AP) is a method of identifying the demographic profiles such as age, gender, location, native language and personality traits an author by processing their written texts. The AP techniques are used in multiple applications literary research, marketing, forensics security. researchers identified various differences authors writing styles analysing datasets. represented stylistic features. extracted several style based features like structural, content, word, character, syntactic, readability semantic to recognize authors. Traditionally, feature combinations for differentiating Several existing works Machine Learning (ML) methods predicting characteristics new author. achieved good accuracies considering both ML algorithms combination. Recently, advent Deep (DL) proposed approaches profiling using these techniques. Few that deep learning performance prediction than results In this work, word embeddings approach gender age prediction. approach, experiment conducted with different embedding models Word2Vec, GloVe, FastText BERT generating vectors words. documents converted document representation technique which uses transferred three Extreme Gradient Boosting (XGBoost), Random Forest (RF) Logistic Regression (LR) trained model. This model predicating accuracy XGBoost classifier other algorithms. implemented on PAN 2014 competition Reviews dataset attained best performances AP.