Black marketed collusive users primary dataset from twitter/x online social media

In the realm of online social media, the proliferation of collusive behavior presents significant challenges for maintaining platform integrity and trust. This study introduces a primary labeled dataset focused on black-marketed collusive users on social media platforms, especially Twitter/X, aiming to classify collusive and genuine social media profiles. Collusive users, often operating in networks to manipulate metrics such as likes, retweets, and followers, were identified through specific patterns of interaction and engagement. Genuine users, on the other hand, were selected based on their organic and non-manipulative activity. The construction of our primary collusion dataset involved a meticulous process of data collection from 4 black marketing sites, followed by extracting features from Twitter/X. This collusive users data was merged with some genuine user data, which were heuristically collected from Twitter/X. Our primary dataset provides a valuable resource for research using machine learning, network science, and social media analysis, enabling the development and testing of algorithms designed to detect colluded users. By facilitating a deeper understanding of collusive dynamics, this work contributes to the broader efforts of safeguarding the authenticity and reliability of social media platforms. This comprehensive dataset will serve as a foundational tool for advancing research in addressing the collusive users Twitter/X social media. For elaborating the possibilities of model building, we have showcased the usage of our dataset with 15 machine learning classifiers, of which the LightGBM model outperformed with an AUC of 0.94. We have also demonstrated model enhancements using hyperparameter optimization with Bayesian Optimizer, Tree-structured Parzen Estimator, and Random Grid Search.