Transactions on Large-Scale Data- and Knowledge-Centered Systems XIII

The traditional search model of finding links on the Web is unsatisfactory for the increasingly complex tasks that seek to leverage the diverse, increasingly structured and semantically annotated data on the Web. A good example is when users seek to find collections or packages of resources that meet some constraints e.g., a collection of learning resources that cover some topics and have a good average rating or a collection of tourist attractions in a city such that total cost and total travel time for visiting all attractions meet the given constraints. For such queries, the goal is the return a set of constraint-qualifying collections or packages. However, using the traditional “set of links” query paradigm, such queries can only be satisfied by issuing multiple queries, reviewing answer lists and manually assembling packages to suit a user’s desired constraints. In this article, we introduce the concept of a Package Query for querying for resource combinations on the Semantic Web. In particular, we consider a frequent subclass of such queries Skyline Package Queries, in which multiple competing criteria are specified in the query so that the pareto-optimal set or skyline of packages are returned. In contrast to a few recent efforts on package queries on single relational models, finegrained data models such as RDF include the challenge of computing the package skyline over multiple joins of ternary relations. We present four evaluation strategies involving different combinations of relational query operators and a new operator for Skyline Package Queries and different storage models for RDF data. A comparative evaluation of the algorithms over real world and synthetic-benchmark RDF datasets is