An Algebra for Path Manipulation in Graph Databases.

A key characteristic of current graph query languages is their support for path queries. Although a path query looks for paths in a graph database, current graph query languages are restricted to return just the source and target nodes connected by each solution path. Therefore, the user is not able to manipulate the elements (nodes and edges) of the resulting paths. In order to overcome such restriction, this paper presents an algebra for path manipulation. Inspired by the relational algebra, we defined the operators of selection, projection, node-based join, edge-based join, node-based cartesian product, edge-based cartesian product, union, intersection and difference. These operators are closed under sets of paths, i.e. the input and the output are sets of paths. We study the algebraic properties of the operators and describe use cases that justify the usefulness of the algebra.