AI and Machine Learning in Supply Chain Optimization: Mapping the Territory

Vishing attacks in financial transactions, the health care industry, and various other domains where sharing of credentials involved using interactive voice response are common. An interactive voice response system (IVR) is an automated mechanism in which the requestors/callers are responded back with the solution to the raised query through prerecorded messages. The objective of this research is to establish a secured IVR system to ensure the traceability options from the customer end through the telephone channels and this is achieved by establishing a secured end-to-end smart contract between the supply chain product traceability system and the customer. The developed model is secured and fortified solution for collecting the information over the voice. Machine learning techniques are applied for detecting the malicious IVR call flow interactions before establishing the smart contract and are carried out by analyzing the nature and type of call request queries raised in the stages of supply chain process. A set of supply chain transactions-based vocabulary is created in analyzing and evaluating the genuineness of the IVR call flows by applying the machine learning algorithm. If the algorithm detects the call flow request as a fake one, then the end user is notified and the establishment of smart contract is blocked. If the algorithm identifies that the call as a genuine one, then the smart contract of the block chain technology is established and a set of secret tokens are exchanged between the caller and the IVR system. If the exchange of tokens is successful, then the interaction commences like business logic and functionalities in supply chain scenarios. On successful completion of the transaction, the life of the generated token is expired. The proposed methodology was tested in the Ethereum platform and the results are promising. In the first phase of the system in detecting the malicious IVR systems, the machine learning algorithm’s success rate was 88%. The generation of exchange of tokens in the second phase of the secured transaction model was a greater level of success at 98.78%. Even if the machine learning allowed the malicious IVR call flow request to proceed to the smart contract, the level of block at this phase was 96.23%..