Resource description framework (RDF) is a data representation standard for loosely structured data that can be used to represent large datasets that arise in applications such as social networks. The challenge of searching and analyzing this data is an ongoing research activity. A number of standards, such as SPARQL, have already been established for querying RDF data, and they provide a standard reference for proposing, comparing, and implementing data analysis techniques and algorithms.

The premise of this paper is that applications typically use multiple queries of datasets. Therefore, the problem query optimization should consider this set of queries instead of optimizing queries individually. The paper presents two heuristic techniques: common subquery identification for rewriting query sets, and a dynamic programming-based optimization approach to sequence “joins” in a query.

The common subquery heuristic is based on clustering sets of SPARQL queries based on their predicates and converting the clusters in a single SPARQL query with OPTIONAL clauses. The result of this query can then be decomposed to obtain results of the individual queries in the original set.

The second technique is based on optimizing the order in which predicates in a query should be evaluated by estimating the number of RDF triples corresponding to each predicate.

The paper is slightly difficult to read because of suboptimal formatting. The example provided is not very illuminating and does not really illustrate the technique of clustering and common subquery optimization. The paper also seems to be quite sparse in terms of references and, in places, has incorrect references. Formal justification for the correctness of the rewriting technique is also missing.