Gender Prediction in Author Profiling Using ReliefF Feature Selection Algorithm

Abstract

Author Profiling is used to predict the demographic profiles like gender, age, location, native language, and educational background of the authors by analyzing their writing styles. The researchers in Author Profiling proposed various set of stylistic features such as character-based, word-based, content-specific, topic-specific, structural, syntactic, and readability features to differentiate the writing styles of the authors. Feature selection is an important step in the Author Profiling approaches to increase the accuracy of profiles of the authors. Feature selection finds the most relevant features for describing the dataset better than the original set of features. This is achieved by removing redundant and irrelevant features according to important criteria of features using feature selection algorithms. In this work, we experimented with a ReliefF feature selection algorithm to identify the important features in the feature set. The experimentation carried on reviews domain for predicting gender by using various combinations of stylistic features. The experimental results show that the set of features identified by the ReliefF feature selection algorithm obtained good accuracy for gender prediction than the original set of features.