The Social-Spammer-Heterogeneous dataset is collected from a social software called Tagged. It contains 5.6 million users and 85 million connections between users. Each user in the dataset contains four features and a manually entered label classification to identify whether it is a spam account. Each connection in the dataset represents an interaction between two users and contains a timestamp and category. There are seven anonymous categories for connections between users in the network. The most original task of this dataset is to identify spam accounts based on associated and unassociated features.