A Provenance Storage Method Based on Parallel Database

When faced with dirty data fused from different sources, modern data management generally trace origins using data provenance technical to repair it fundamentally. Unfortunately, the state-of-art data provenance approaches can only deal with small amount of data using annotation or inverse process. What's worse, these approaches just work under stand-alone mode, resulting in low efficiencies. In this paper, we raise a distributed provenance storage strategy DPSM under the environment of parallel data-base combined with the small-amount-storage advantage of "Provenance Tree". The model optimizes both query efficiencies and storage cost by using a detached method to store different provenance information.