Using Social Relationships for Computing Reputation Clause Samples

Using Social Relationships for Computing Reputation. ‌ In the work done by ▇▇▇▇▇▇▇ and ▇▇▇▇▇▇ [246], social network analysis, which is based on studying the social relationships between peers, is used to provide an additional source of information (to traditional information obtained from direct interactions and those obtained from other members of the society about their past experiences) for computing reputations. The basic idea of the system is that reputation is based on three dimensions: the individual dimension, the social dimension, and the ontological dimension. The individual dimension models the direct interactions between two agents. In this dimension, the subjective reputation is calculated directly from an agent’s im- pressions database. However, the reputation value on its own is not sufficient. Hence a reliability measure of this value is also calculated by taking into account the number of impressions used to calculate the reputation value and the variability of their rating values. The social dimension models the case when the source of information comes from other agents in the system. Depending on the information source, three types of social reputations arise: witness reputation, neighbourhood reputation, and system reputation. In witness reputation, to identify witnesses, first the most connected sub-graphs are obtained, then the nodes with local centrality are identified. This way the peer will end up using only the most representative agents for its source of information; hence, it will be minimizing the correlated evidence problem. The next step would be to aggregate all this witness information. The final reputation measure would be an aggregation of the reputation values provided by all witnesses, taking into account how much trustworthy is each witness in providing its reputation value of the target agent. The final reliability measure is also an aggregation of the reliability and trust measures of each individual witness. The trust measure is defined as follows. The degree of trust that an agent a has in agent b on providing feedback about agent c is a combination of subjective trust reputations, which are calculated in a similar manner to individual reputation, and social trust. As for social trust measures, these are the results of applying fuzzy rules. These fuzzy rules depend on the type of relationships in a specific scenario, such as competitive, cooperative, and trade relationships. Note that different fuzzy rules would exist for different contexts and scenarios. The basic ...