INTEGRATING CMP-BLOCKCHAIN WITH NLP AND MACHINE LEARNING FOR TRUST VERIFICATION AND EVENT DETECTION

Abstract

An integral part of our everyday lives is spent on social networks, and one of the most important parts of these networks is the so-called social reviewing system (SRS), which allows us to access data, usually in the form of reviews. The significance of social networks necessitates that they be trustworthy and secure, preventing assaults and misuses and allowing users to freely utilise the information they provide. False reviews are a major weapon in the fight against the reputation system. Since even verified members of the network are capable of launching such attacks, a strong defence is to take advantage of trust management by giving each user a trust level and then having them use it to evaluate the collected data. Because it is subjective and difficult to completely automate the process of detecting improper behaviours, trust management within the framework of SRSs is especially complex. Despite several proposals in the existing literature, this matter has not yet been fully addressed. By using the innovative notion of time-dependent and content-dependent crown consensus and modelling trust management as a multicriteria multiexpert decision making, this work proposes a remedy against mendacious reviews that integrates fuzzy logic with the theory of evidence. Even when faced with sockpuppet assaults, our method proved to be more effective than the primary methodologies described in related literature.