Improving the performance of recommender systems based on blockchain technology

Abstract

Blockchain technology has received the focus of many scientists as a promising technology in distributed systems. It was made possible by blockchain aspects such as visibility and Permanence. The newness of blockchain technology causes many challenges in this field. One of these challenges is managing data in the blockchain and presenting data suitable to the user's interests. In current centralized systems, recommender systems have solved this challenge. Implementing recommender systems in blockchain intelligent contracts and raising the transaction cost make the recommendations inaccurate due to the lack of complex calculation facilities of machine learning algorithms in innovative contract programming languages. They were introducing a novel method for enhancing recommender systems using blockchain technology. This method involves storing data in the blockchain according to a structure stipulated within the smart contract. The data is then provided to the off-chain recommender system through the public key of the smart contract, enabling the necessary processing for providing the appropriate recommendation to the user. The results are then stored in the blockchain and presented to the user during a transaction. The results of this method, as compared to previous research and works, demonstrate that performing complex calculations outside the chain not only reduces the transaction cost of establishing a contract but also decreases the transaction cost related to the recommender in the proposal system in terms of gas consumption, leading to increased scalability.