Data Provenance Management for Fine-grained and Coarse-grained levels

Abstract

Data Provenance provides a semantic mean about the input contributed to the existence of a query output; and how data evolves over the time. Hence data provenance has an excessive usage in gaining the accuracy and trustworthiness of data that users query. User can query data from base tables, curated tables and table views. The curated tables and views may have been created using other base tables, curated tables and table views. Here the proposed provenance management system shows how a tuple in a query output was derived from the original base tuples by going through the curated tables and views up to the base relation level. System rewrites user select query using several rewritten rules in order to generate the input data that contributes to the query output. Since there is a need of maintaining the correct reference between the output tuple and contributed input tuple(s) both query output and the provenance data are represented in the same output. While the above scenario is considered as the coarse-grained level the ability to show the evolvement of data attribute values over the time is provided in the ne-grained level. The user selected, provenance needed attributes are separately maintained in the provenance data store in order to capture their evolvement over time. The main objective of the suggested approach is not to alter the user databases by changing or adding provenance related tables or values. Hence a separate data store is maintained for provenance data management. Here the experiments on provenance management process indicate that the suggested approach e ciently provides provenance information inducing only a small overhead on queries.