The broader impact/commercial potential of this I-Corps project (PaySpray) is to provide a platform in which the members offer their products or services to the network, increase their assets, and use them to reduce their debts. Each member of the system is a network node who is connected to the other members through financial links that represent debt or credit relationships. Once a transaction occurs between two members of a network, either inside a system or outside of it, the link including transaction information will be created between them. PaySpray owns the knowledge over all the networks supplies, demands, and transactions and finds the optimum way to suggest the services from members to members to close a cycle. Members can be individuals or companies, and services can be anything legal to be sold or done. A member of a cycle may be a member of multiple cycles. Services may exceed the amount of the owed money or be less than that, the remaining debt can be paid either by paying money or providing another service or entering into a new cycle. The object is to create a network that permits various type of payments, bartering and financial transactions to manage an individuals or organizations finances.
This I-Corps project utilizes Artificial intelligence on a platform that intelligently connects users for financial transactions. Members may invite anyone to join their network that they have financial relationships with whether they are the payee or recipient of the finances. This process continues until until a cycle is created. The system permits multiple cycles, and members can participate in multiple cycles.
In developing the application, a heuristic complex network algorithm was utilized to identify the possible loops, matching service and demand in the networks. This was also used as a means to improve the efficiency and optimize performance of the model. The research also applied machine learning algorithms (mainly Natural Language Processing) to extract the information from users, inputs and use the output of the ML model to match service and demands. As a future improvement, researchers will apply deep learning RNN models in order to detect the fraud services and demands in the app. Currently the application is developed in AngularJS with MongoDB Databases and deployed on Amazon Web Service.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.