Towards A Machine Learning Enabled Multi-Channel Messaging Framework for Financial Service Institutions: Preliminary Investigations

摘要

Messaging is essential when organizations dealing with financial customers need to share information. Technological innovations, such as machine learning (ML), have provided financial service institutions (FSIs) with the ability to reach out to consumers in more intelligent ways. Multi-Channel Messaging System (MCM), in use by FSI's, enables the seamless integration of disparate channels of communication within a single system. This problem has driven us to develop a machine learning-enabled channel assignment algorithm in the Multi-Channel Messaging System. The decision-making module would incorporate heterogeneous channels using an Enterprise Service Bus (ESB) layer, and use machine learning algorithms to assess channel capacity, dynamic assignment, and customer trends. Our framework would extend and maximize this approach, to achieve a fast balance between the exploration-exploitation dilemma, for channel selection. This research explores the problems and challenges a multi-channel messaging system currently used by financial services organizations, while proposing a multichannel framework which incorporates a machine learning channel selection and learning module. The analysis concentrates on the complex approach to channel selection and integration methods that can make the system effective and usable. The lessons learned from the design would be further refined to inspire future work in this field.