The dynamically increasing trend of the regulation requirements and the succeeding intensification of the amount of transactional and communication data have posed critical questions with the observation of the conformity inside the companies. Financial institutions, medical care providers, and other institutions that are under control are also under pressure to identify suspect actions and decrease compliance expenses and efficiency. Although it might seem that old rule-based systems of monitoring are not so young anymore, they are getting less efficient to combat these challenges. These systems are based on programmed limits, rules and keyword recognition which tend to produce too many false positives which produce operational inefficiencies and regulatory risks. Possible alternatives can be provided by the recent developments in natural language processing (NLP) and semantic analysis. These procedures assist systems to perform linguistic context, intent and unstructured data examination in a superior way. The majority of the latest advances in word embeddings, transformer models that are based on domain-specific financial and compliance purposes and applications have triggered access to contextual and scalable monitoring. NLP could offer deeper meaning of messages and transactions under the element of going beyond the concept of identifying keywords and ultimately narrowing down the number of false alarms and enhancing effectiveness of compliance control. In this paper, monitoring architecture is suggested, and it includes a group of transactional and communication streams of information along with a combination of artificial intelligence (AI) and machine learning (ML). To be exact, semantic role labeling, knowledge graphs, and transformer embeddings are used to offer further opportunities to detect and lessen the load of compliance departments. A part of the work of the research is provided in the developed system architecture, critical survey of the current methods, and discussion of the application in the industry. The case study of HSBC, Dynamic Risk Assessment program, and JPMorgan, COiN system shows how powerful AI-compliance systems are revolutionary.
Pratik Chawande (Wed,) studied this question.