Trust computation in web services based on Beth's trust model

Abstract

Security mechanism is obviously vital in any systems especially when it comes to web services. To protect web resources from malicious attacks, trust relation is required between entities. Trust relation between entities was a key factor for the successful rate of web operation. Against the problem of failing to punish malicious recommendation, Beth's trust model was improved by introducing a punishment factor; it makes sure that the transaction behaviours can be reflected by trust degree so as to eliminate malicious recommendations. In this way the trust model is more reasonable and the trust value is more subjective. Experiment results show that the trust value of an entity increased according to more successful transactions, which embodies the purpose of giving high-trust-degree entity priority to access services and punishing malicious entities. When malicious entity's proportion is large, a higher weight factor φ can guarantee successful transaction rate via better recognising malicious entities.