Predicting User Engagement in Social Networks using Machine Learning withCNN Algorithm

In recent years, social network platforms have become increasingly popular as a means of communication and marketing. Twitter, in particular, generates a vast amount of user-generated content that can be used to predict user engagement. In this study, we propose a novel approach to predicting user engagement with Twitter content by combining text and image data using Convolutional Neural Networks (CNN). Dataset is Twitter Dataset for Sentiment Analysis .In addition, the dataset includes an images folder with subfolders for positive, negative, and neutral images and a LabeledText.xlsx file. We preprocess the text and image data separately and use a multi-input CNN model to predict the engagement metrics. Our model achieves an accuracy of ۶۸% on the validation set, demonstrating its ability to predict user engagement with Twitter content using both text and image data. We also perform a feature analysis to identify the most important words and image features that contribute to user engagement. Overall, our study shows that combining text and image data can improve the accuracy of user engagement prediction models and provides insights into the factors that influence user engagement on social media. This information can be valuable for social media marketers and content creators to optimize their content for maximum engagement.

کلیدواژه ها:

user engagement ، CNN ، accuracy ، text data ، image data ، multi-input model