Analysis of the performance of reputation query in different frameworks

Reputation of the entities is very important for other entities to evaluate the trustworthiness of them. This paper analyzes the efficiency of the query in different frameworks using differential methods. In centralized framework, there is a centre to store reputations and provide them to other entities, and the number of queries is only once for every query, so the efficiency is very high, but it has poor expansibility; in the pure decentralized framework, it´s convenient to store reputation in the network, but it´s hard to query exhaustively and accurately, and query messages will increase system cost; in partial decentralized framework, the network is divided into some reputation domains, and the performance is analyzed from inside domain and outside domain, and this framework has high efficiency, low system cost (involved 2 or 3 entities in one transaction) and message cost (2 or 4 messages in one transaction), and the results show that the query performance will be much better if the entities, which have more transactions, are divided into a same domain.