Prediction of the Engagement Rate on Algerian Dialect Facebook Pages

For the purposes of online marketing, some social networks provide an advertising platform that allows the sponsoring of advertising content to reach target users. This content promotion is expensive in terms of the budget to be spent and this is why the content to be sponsored must be carefully selected. In other words, a company would ideally only sponsor content which is likely to perform well. The performance of an advertising content is usually measured by a metric called the Engagement Rate often used in the field of social media marketing to measure the extent to which the users will show “interest” for and interact with the advertised content. Thus, being able to predict the engagement rate of a publication is of utmost importance to Social Marketers. In this work, we propose a deep-learning-based system, to predict the performance of Facebook posts content in the Algerian Dialect, as measured by the users’ engagement rate with respect to these publications. In order to predict the engagement rate, the system processes all the publication content: the text, the images, and videos if they exist. The images are preprocessed to extract their features and the Algerian Dialect textual content of the posts is analyzed despite its complexity which is due to multilingualism (use of Arabic, Algerian dialect, French and English). Two models of neural networks were proposed, one based on an MLP architecture and the other on a hybrid Convolutional-LSTM and MLP architecture. The results produced by these models on the prediction of the engagement rate are compared and discussed.