Text this: Network Representation Based on the Joint Learning of Three Feature Views