In e-commerce systems like online auction houses or online stores, there are financial transactions involving buyers and sellers. At large payment processing firms, there is significant risk of fraud (upto 0.9 %). This fraud can be prevented before the actual transaction phase through risk scoring models. In the post transactions phase, measures like withholding or reserving funds of the seller, or asking for additional supporting material from the seller to release the funds can be done. There are numerous variations based on different geographies or different seller classes or different holding mechanisms of these measures.
It was found that the software being developed for a payment system was combining both infrastructural software (database, queue, logs) as well the actual risk business process. Subsequently, prototypes were created for different risk measures in the post transaction phase using executable BPMN2 (using Activiti engine). For example, a certain amount of money may be withheld from the seller for a configurable time period which can be edited in the graphical BPMN2.
In this paper, we discuss the numerous types of transactions, the numerous measures for financial risk and how BPMN2, can be used to model the same and at the same time form a performant executable component reducing development time. It is also possible that such models can be standardized and exchanged in the industry.