Design of Declarative Graph Query Languages: On the Choice between Value, Pattern and Object Based Representations for Graphs

Graphs play a major role in many applications in business, science, and social sciences allowing modeling and analysis of interactions among objects and their structural relationships. The recent push for analytical methods to study inter-object associations in these domains has resulted in many interesting algorithms such as page rank and random walk, and systems such as TALE, SUMMA, QuickSI and C-Tree. While these algorithms address different aspects of graph analysis with varying degrees of effectiveness, their focus is not on declarative querying of graphs except GraphQL and IsoSearch. However, the first-order framework on which both GraphQL and IsoSearch are based does not allow for many queries of interest. In this paper, we argue that this limitation may be, at least partially, removed by implementing novel graph operators for relational databases. We present the idea of minimum hub covers as a basis for a compact and non-redundant representation of graphs in nested relational databases, and leverage this representation, called g-relations, toward graph operator implementation for graph isomorphisms and graph comparisons. This is an ongoing research and hence, depicts partial results.