Market-based SPARQL brokerage with MaTriX: towards a mechanism for economic welfare growth and incentives for free data provision in the Web of Data

The exponential growth of the Web of Linked Data (WoD) has so far primarily been funded using subsidies, where new datasets are financed through public funding or via research programs. Relying on (public) subsidies, however, may eventually limit the growth of the WoD, focus on areas decided by committee rather than true demand, and could hamper data quality due to the lack of clear incentives to maintain high quality standards. In this paper we propose a market-based SPARQL broker over a heterogenous, federated WoD as a economically viable growth option. Similar to others, we associate each query with a given (potentially zero) budget and a minimal results-set quality constraint. The SPARQL broker then employs auction mechanisms to find a desirable set of data providers that jointly deliver the results. We evaluate our market-based SPARQL broker called MaTriX using a simulation. Our results show that a mixture of free and commercial providers actually provide superior performance in terms of consumer surplus, producer profit, total welfare, and recall whilst being incentive compatible with the provision of high-quality results. We even found that the increase of profit in the mixed situation may entice commercial providers to subsidize free providers directly.