Provider issues in qualityconstrained data provisioning

Formal frameworks exist that allow service providers and users to negotiate the quality of a service. While these agreements usually include non-functional service properties, the quality of the information offered by a provider is neglected. Yet, in important application scenarios, notably in those based on the Service-Oriented computing paradigm, the out- come of complex workflows is directly affected by the quality of the data involved. In this paper, we propose a model for formal data quality agreements between data providers and data consumers, and analyze its feasibility by showing how a provider may take data quality constraints into account as part of its data provisioning process. Our analysis of the technical issues involved suggests that this is a complex problem in general, although satisfactory algorithmic and architectural solutions can be found under certain assumptions. To support this claim, we describe an algorithm for dealing with constraints on the completeness of a query result with respect to a reference data source, and outline an initial provider architecture for managing more general data quality constraints.