A Method of Combining Hidden Markov Model and Convolutional Neural Network for the 5G RCS Message Filtering

As one of the 5G applications, rich communication suite (RCS), known as next generation Short Message Service (SMS), contains multimedia and interactive information for a better user experience. Meanwhile, RCS industry worries that spammers may migrate their spamming misdeeds to messages, complexity which challenges filtering technology because each them hundreds fields with various types data, such texts, images videos. Among text data contain main content, is adequate more efficient combating spam. This paper first discusses fields, possibly spam information, then use hidden Markov model (HMM) weight finally convolutional neural network (CNN) classify messages. In HMM step, are treated differently. The short texts these represented feature sequences extracted by extraction algorithm based on probability density function. Then, proposed learns sequence produces proper text. Other fewer words also weighted algorithm. CNN all weights construct message matrix. matrices training messages used inputs learning testing trained property prediction. Four optimization technologies introduced into classification process. Promising experiment results achieved real industrial data.