Page 83 - Trust in ICT 2017
P. 83
Trust in ICT 1
8.4.3 Trust Models and Trust Metrics
Based on the approach mentioned in the Trust Model in previous sub-section, with the catalyst of imitating
human trust processing as discussed above, a trust model comprises of three TMs namely Reputation,
Recommendation, and Knowledge (See Figure 24).
Figure 24 – A Trust Model with three Trust Metrics
This sub-section takes the trust-car sharing example for illustrating the policy mechanism reasoner.
Generally, the Reputation and Recommendation TMs in the trust car-sharing example are similar to any other
services; and can be get from the reputation system. The Human-to-Human knowledge can be also calculated
depending on four TAs mentioned in the previous section. The Human-to-Object knowledge extraction
algorithm and Trust Calculation mechanism are service-and-object specific.
Knowledge is the first party information provided by trustee to evaluate its trustworthiness and composed
by some TAs depending on services and entities. Service providers are supposed to register their own
information including both Knowledge TM ontology and requirements to the platform prior to use. These
trust data has many dimensions and should be normalized and unified in order to be suitable for software
oriented architecture (SOA) environment by using an ontology manager and an information model.
This report considers the platform for social IoT environment in which humans offer services through their
owned items. Thus, when judging Knowledge TM of a service, a user needs to assess both device and device’s
owner as illustrated in Figure 25.
The Human-to-Human knowledge can be comprised of four TAs: Honesty, Cooperative, Community-Interest
and Experience, inspired by ideas in [84].
Figure 25 – The Knowledge TM is divided into two sub-ontologies
The honesty represents whether a human is honest. In social IoT, a malicious user can be dishonest when
providing services or trust recommendations, resulting in disrupting the trust management and service
continuity. Thus, honesty is chosen as a TA to prevent an entity from trusted-related attacks.
75