A Framework for Fine-grained Data Integration and Curation, with Provenance, in a Dataspace

Some tasks in a dataspace (a loose collection of heterogeneous data sources) require integration of fine-grained data from diverse sources. This work is often done by end users knowledgeable about the domain, who copy-and-paste data into a spreadsheet or other existing application. Inspired by this kind of work, in this paper we define a data curation setting characterized by data that are explicitly selected, copied, and then pasted into a target dataset where they can be confirmed or replaced. Rows and columns in the target may also be combined, for example, when redundant. Each of these actions is an integration decision, often of high quality, that when taken together comprise the provenance of a data value in the target. In this paper, we define a conceptual model for data and provenance for these user actions, and we show how questions about data provenance can be answered. We note that our model can be used in automated data curation as well as in a setting with the manual acti...

• Type: Journal
• Year: 2009
• Where: FAST
• Authors: David W. Archer, Lois M. L. Delcambre, David Maier
• Keywords: Data Curation, FAST 2009, Heterogeneous Data Sources, Operating System, Provenance