SemantiChain: A Trust Retrieval Blockchain Based on Semantic Sharding
Zihang Zhen,Xiaoding Wang,Xu Yang,Jiwu Shu,Jia Hu,Hui Lin,Xun Yi
DOI: https://doi.org/10.1109/tifs.2024.3488501
IF: 7.231
2024-11-19
IEEE Transactions on Information Forensics and Security
Abstract:Since its inception, blockchain technology has found wide-ranging applications in various fields including agriculture, energy, and so on, owing to its immutable and decentralized nature. However, existing blockchains encounter significant challenges in scenarios that demand efficient retrieval of big data. This is primarily because current blockchains cannot directly store and process diverse types of rich media information. Additionally, the semantic relationships between data within the blockchains are weak, complicating the categorization and retrieval of data and transactions. Moreover, the scalability of current blockchains is limited, with the capacity of full nodes continually increasing. Although some semantic-based blockchain solutions that combine off-chain scalability have been proposed, they are limited in effectiveness and applications. To address these issues, this paper introduces a brand-new blockchain sharding technique called Semantic Sharding, which enhances blockchain scalability through a hybrid on/off-chain approach. Building on this, we propose a semantic sharding blockchain architecture, SemantiChain, which enables the on-chain storage and retrieval of transaction semantic features. Furthermore, through the Po2RW consensus protocol, we balance the scalability and security of SemantiChain. Security analysis proves that SemantiChain can resist security risks such as man-in-the-middle attacks, malicious node attacks and on/off-chain data inconsistency. Experimental results demonstrate that SemantiChain can reduce search time and memory usage by at least 32.29% and 77.97% respectively under the same retrieval performance, compared to mainstream approximate nearest neighbour retrieval algorithms. Furthermore, compared to the SOTA semantic blockchain, SemantiChain achieves a retrieval performance improvement of at least 45.88% and reduces retrieval memory usage by 95.76%.
computer science, theory & methods,engineering, electrical & electronic