Source Attribution for Querying Against Semi-structured Documents.

Abstract

Mediation architectures like the Context Interchange re-search project, from which this work stems, integrate dis-parate information sources, hiding distribution and recon-ciling heterogeneity. As a result of transparent access, the notion of distinct sources often disappears from queries and results. However, there are situations where users or ap-plications do need to know the sources from which a par-ticular datum is drawn: for example, enforcement of intel-lectual property, evaluation of data quality or measurement of the timeliness of data. In this paper, we define attribu-tion as the association of a value in the result of a query with the sources either from which the data was extracted or which contributed to the selection. Motivated by multi-source querying, attribution has particular applicability in the context of the World Wide Web. After beginning with an example that both describes and motivates the need for attribution within the semi-structured environment of the Web, we offer a general attribution framework and algebra and discuss the implementation of a prototype. 1