Browse Theses

In commercial network database management systems, set-valued fields and aggregate functions are commonly supported. However, the relational database model, as defined by Codd, does not include sex-valued attributes or aggregate functions. Recently, Klug extended the relational model by incorporating aggregate functions and by defining relational algebra and calculus languages.

In this thesis, relational calculus database query language (as defined by Klug) is extended to manipulate set-valued attributes and to utilize aggregate functions. The expressive power of the extended language is shown to be equivalent to the extended relational algebra (ERA) of Ozsoyoglu and Ozsoyoglu which includes three new operators, namely, pack, unpack and aggregation-by-template. The extended languages form a theoretical framework for statistical database query languages.

Summary-Table-by-Example (STBE) is a graphical, user friendly language based in the extended relational calculus. STBE, suitable for statistical database applications, permits queries with a hierarchical subquery structure, and manipulates relations with set-valued attributes and summary tables.

The hierarchical arrangement of STBE queries naturally implies a tuple-by-tuple subquery evaluation strategy (similar to the nested loops join implementation technique) which may not be the best query processing strategy. In this thesis we discuss the query processing techniques used in STBE. We first convert an STBE query into an extended relational algebra expression using techniques similar to those proposed for removing the nesting from SQL queries. Two transformations are introduced to remove the hierarchical arrangement of subqueries so that query optimization is possible. To solve the "empty partition" problem of aggregate function evaluation, directional join (one-sided outer-join) is utilized. We then give the algebraic properties of the ERA operators to obtain an "improved" ERA expression. Finally we list alternative access paths and their cost formulas for obtaining an access path with the smallest cost. In addition to revising the access paths from SQL and ABE (Aggregates-By-Example) for STBE, new access paths for the ERA operators pack, unpack, and the aggregate-by-template are presented.