Abstract

This study focuses on managing trust in information sharing in agricultural sector. Awareness of smart farming technology in the agricultural industry for information dissemination among farmers remains low. Blockchain is a suitable solution for trust issues in information sharing due to its decentralized, transparent, and immutable nature, and it has evolved from cryptocurrency to other applications through Ethereum. Various blockchain consensus mechanisms have been developed to address performance challenges. Some of the commonly used consensus mechanisms for blockchain technology include Proof of Work (PoW), Proof of Stake (PoS), Proof of Reputation
(PoR), Proof of Trust (PoT), and Proof of Activity (PoAc). Existing consensus mechanisms select nodes based on a scoring system, followed by random selection. In this study, machine learning is used to replace the selection method using N-Gram model to predict scores. The N-gram model is introduced to replace the random node selection method. An N-gram model groups sequences of words, counts how often each sequence appears, evaluates predictions against actual data, and calculates the probability of each sequence by dividing its frequency by the total number of Ngrams. Choose the prediction score with the highest probability based on the matching selection. Tests were conducted on all commonly used consensus mechanisms by evaluating transaction speed per second, block creation per second, and energy consumption, with results of 10 transactions per second, 13 blocks created per second, and 10 units of energy consumed, which are better than those of other blockchain consensus mechanisms.