WEB PAGE USER INTERACTION ON SOCIAL MEDIA

The study explores the usefulness of incorporating both image and text data to predict user engagement on social media. The research collected *350k Facebook posts* and incorporated five types of machine learning models: text-based Neural Networks (NN), image-based Convolutional Neural Networks (CNN), Word2Vec, decision trees, and a combination of text-based NN and imagebased CNN. The models are unique in their use of the data. The models learn and test on advertisement posts in order to predict user engagement, which includes share count, comment count, and comment sentiment. The study found that *combining image and text data produced the best models*. The research further demonstrates that combined models outperform random models.