Trustworthiness Assessment of Knowledge on the Semantic Sensor Web by Provenance Integration

Knowledge represented on the Semantic Sensor Web originates from different datasets which are often a collection or aggregation of other sources. The SSW is dynamic, open and distributed, so the datasets are of varying quality and completeness. Consumers need to be provided with a level of trustworthiness of this knowledge to determine its relevance and usefulness. Interpretation of provenance (detailed information about the origin of data - held in metadata) is necessary in order to analyse how knowledge came into existence and measure its trustworthiness. However there are challenges in interpreting the provenance in a uniform way, because different data providers use different processes to manipulate the data and different annotation techniques to provide metadata. Although there are methods for retrieving provenance, knowledge consumers are left with the responsibility of assessing the trustworthiness of discovered knowledge dependent on how they see it fitting their application. This paper proposes a meta-knowledge ontology to align the concepts and properties of existing provenance schemas and ontologies. The meta-provenance ontology enables common interpretation of different provenances, and hence their integration. This paper also presents a trustworthiness assessment model based on integrating provenance. This model provides a function for the knowledge consumer to choose the relevant provenance attributes and allows for ranking of their importance. This provides a reliable mechanism for measuring trustworthiness, as only attributes relevant to the consumer are used.