The goal is to create a model for forensic authorship attribution of online social text using character n-gram feature with several convolutional filters integrated with bidirectional LSTM to understand the morphological, thematic content and style of individual authors.