A Design For A Machine-Learning-Enabled Multi-Channel Messaging Framework for Financial Service Institutions

Financial Services Institutions (FSIs) utilize multi-channel messaging systems (MCM) to integrate diverse consumer communications channels into a unified system for the benefit of their customers. However, the multi-channel service has an extra overhead coordinating channel selection to identify which channel is available to send messages to customers because of the disparate nature of the integrated channels. Machine learning can be used to determine channels for communication in a customer alert system used by financial services institutions (FSIs). This work offers a machine-learning-enabled channel selection and management method based on Tug Of War (TOW) algorithm. The framework proposed will be based on the TOW dynamics algorithm. TOW uses a modified learning technique based on addition and subtraction's reward/loss principle. Furthermore, we demonstrate the efficacy of this strategy by prototyping the proposed framework in a customer alert system termed the ML-Enabled MCM system. Messaging channels such as SMS, Email, Twitter Direct Messages, Instagram Direct Messages, Telegram, and Facebook Direct Messages were integrated with an Enterprise Service Bus (ESB) responsible for channel management. The new design reveals that the machine learning-enabled MCM system can learn and adapt quickly in response to channel feedback and make quick decisions about which channels a customer can receive messages on while maintaining balance within the system. The proposed framework will be relevant to the event and message processing products offered by FSIs.