Efforts to comprehensively describe behavior have advanced representation-based methods, particularly those using machine intelligence models like deep learning models, to capture behavioral information. As behavior complexity increases, the expressive power of these intelligence models reaches a bottleneck. Behavior possesses inherent structural information, which allows for efficient description of increasingly complex behaviors. To concretize behavior structure into a format suitable for processing, this book adopts graphs to reflect the relationships between behavioral attributes at a fine-grained level and designs the behavioral molecular structure model. This book designs the different representative downstream tasks to test the performance of the model on renowned public datasets. The model has high performance in detecting behavior, predicting behavior, and generating behavior. Overall, computational structural behavior is a framework that deciphers behavioral patterns with precision akin to engineering analysis at the atomic level and offers a high-performance and expressive approach for behavioral computing research.