Sideway Value Algebra for Object-Relational Databases

Using functions in various forms, recent database publications have assigned “scores”, “preference values”, and “probabilistic values” to object-relational database tuples. We generalize these functions and their evaluations as sideway functions and sideway values, respectively. Sideway values represent the advices (recommendations) of data creators or preferences of users, and are employed for the purposes of ranking query outputs and limiting output sizes during query evaluation as well as for application-dependent querying. This paper introduces SQL extensions and a sideway value algebra (SVA) for object-relational databases. SVA operators modify and propagate sideway values of base relations in automated and generic ways. We define the SVA join, and a recursive SVA closure operator, called TClosure. Output tuples of the SVA join operator are assigned sideway values on the basis of the sideway values and similarities of joined tuples, and the operator returns the highest ranking tuples. TClosure operator recursively expands a given set of objects (as tuples) according to a given regular expression of relationship types, and derives sideway values for the set of newly reached objects. We present evaluation algorithms for SVA join and TClosure operators, and report experimental results on the performance of the operators using the DBLP Bibliography data and synthetic data.