Similarity Matching, Classification, and Recognition Mechanism for Transaction Analysis in Blockchain Environment

Abstract

The electronic era is providing many opportunities to the consumers by exploiting advanced technologies such as Blockchain, AI driven solutions, Internet of Everything (IoE), optical technologies and 6G networks. Blockchain users need to analyze and examine the diverse transactions recorded on the ledger. In Blockchain technology, transparency and decentralization are the key features for the users and transaction analysis plays integral role in ensuring the privacy of data, security of user credentials, and efficiency of the Blockchain network. By analyzing the features of data, patterns of data, and behaviors of transactions, the Blockchain users can determine the flow of digital assets and also examine the pulse of the Blockchain ecosystem. The smart contacts in blockchain are also seeking for the transaction analysis as it is very important for implementation of the smart contracts to follow transparency attributes and transaction conditions along with patterns in transactions. This research is using two-fold methodology. In first module, a double-layer model (DLM) is designed. The DLM is constructed through the target detection layer and target segmentation layer. The experimental outcome shows that the proposed algorithm has high classification accuracy of 94%. In second module, a Cosine similarity is used to find distance in the data-points and then the XGBoost classifier classifies the transaction data more easily by attainting the classification accuracy of 98% with lower time complexity.