Project overview

JASPR enables the use of provenance records to provide personalised and rationale-backed reputation assessments by utilising an architecture in which a client makes requests to an assessor for reputation assessments (please refer to [1] for more details). The assessor relies on provenance graphs, rather than a list of independent interactions, to make its assessments. Provenance records are recorded as a side-effect of interactions, by one or multiple parties in prior interactions, providing crucial evidence that may be missing for assessing reputation. For example in a logistics chain, in addition to clients recording information, each provider can record information about the sub-contractors that are used, giving the assessor detailed information about sub-contractors’ performance in a range of contexts.

This allows mitigation, situation, indirect responsibility, and other such context to be accounted for, and for the interdependencies of providers to be understood. Mitigation can have many forms, such as a subsequently replaced sub-contractor failing to deliver on time or a client failing to specify all the required conditions (e.g. expiration date of goods being shipped). 

We look for patterns in the provenance that indicate situations relevant to the current client’s needs and mitigating circumstances, and filters the provenance for key subgraphs from which reputation can be assessed. The subgraphs give a description of what has occurred, allowing the reputation measure or recommendation returned to the client to be accompanied by a rationale. 

Further, we aim to personalise reputation assessments. Clients decide which service to interact with based on the reputation or recommendation given, plus the rationale, and on their own preferences and reasoning. The interaction due to this choice, and its link back to the assessment provided by the assessor, is itself included as part of the provenance records. This helps distinguishing which situations are actually relevant to the particular client.

[1] N. Griffiths and S. Miles, An Architecture for Justified Assessments of Service Provider Reputation, in Proceedings of the 10th IEEE International Conference on e-Business Engineering (ICEBE 2013), pp. 345-352, 2013.