Gender Prediction of Journalists from Writing Style

Web-based Kurdish media have seen a tangible growth in the last few years. There are many factors that have contributed into this rapid growth. These include an easy access to the internet connection, the low price of electronic gadgets and pervasive usage of social networking. The swift development of the Kurdish web-based media imposes new challenges that need to be addressed. For example, a newspaper article published online possesses properties such as author name, gender, age, and nationality among others. Determining one or more of these properties, when ambiguity arises, using computers is an important open research area. In this study the journalist’s gender in web-based Kurdish media determined using computational linguistic and text mining techniques. 75 web-based Kurdish articles used to train artificial model designed to determine the gender of journalists in web-based Kurdish media. Articles were downloaded from four different well known web-based Kurdish newspapers. 61 features were extracted from each article; these features are distinct in discriminating between genders. The Multi-Layer Perceptron (MLP) artificial neural network is used as a classification technique and the accuracy received were 76%.