Trust Management Systems often make use of evidences such as certificates or users? recommendations that are part of a separated infrastructure added to applications for trust management purpose. Alternatively, trust computation might be based on evidences gathered directly from core application elements. Our ongoing research focuses on the latter approach proposing a methodology for selecting trust evidences among internal elements of the application under analysis. Key features of our method are its generality, the possibility to offer human-based justifications and the usage of presumptive reasoning to ground its decisions.

If successfully identified, trust evidence could support a trust-based computation that is application-contained and non-invasive, compatible with aspects of ubiquitous and pervasive scenarios. In our evaluation section we show a complete example of our methodology.