SciProv: uma arquitetura para a busca semântica em metadados de proveniência no contexto de e-Science

Abstract

E-Science is characterized by manipulation of huge data set and large scale computing resources usage, often located in distributed environments. In this scenario, represented by high complexity and heterogeneity, it becomes important to treat data provenance, which aims to describe data that were generated during a scientific experiment execution and presents processes of transformation by which underwent. Thus, lineage helps to form a quality, validity and topicality vision of data produced in a scientific research environment. SciProv consists of an architecture that aims to interact with scientific workflows management systems for capture and manipulation of generated provenance metadata. For this purpose, SciProv adopts an approach based on an abstract model for representing the lineage. This model, called Open Provenance Model, provides to SciProv the ability to set up a homogeneous and interoperable infrastructure for handling provenance metadata. As a result, SciProv is able to provide a framework for query data provenance generated in a complex and diverse e-Science scenario. More important, the architecture makes use of semantic web technology to process metadata provenance queries. In this context, using ontologies and inference engines, SciProv provides resources to make inferences about lineage and to obtain important results in allowing the extraction of information beyond those that are registered explicitly from managed data.