Forschungsthemen
[MA] A Semi-Automatic Transformation from Object-Oriented Systems to Role-Based Systems
Motivation
The research training group RoSI investigates advances in role-based information systems. Especially, it extends common data models with new features, like Roles, Relationships, and Compartments, in order to capture dynamic, complex applications domains. In particular, it introduces a new conceptual modeling language, Compartment Role Object Model (CROM) accompanied by a conceptual query language called RSQL.
The latter language, is used to define, manipulate and query an applications role-based data model. Moreover, there exist a prototypical Database Management System (DBMS) supporting the core concepts, namely Objects, Roles, and Relationships. While such a DBMS can naturally capture the semantics of CROM, as seen below, its performance must be evaluated in a comparable way.
SELECT p2.name, a.balance
FROM Person p1 PLAYING Consultant co,
Person p2 PLAYING Customer c,
Account a PLAYING CheckingsAccount ca,
RELATING co WITH c USING advices ad,
RELATING c WITH ca USING owns o
WHERE ca.limit > 1000 AND p1.name="Peter";
Problem Definition
However, there exists no benchmark for role-based information systems incorporating Role Types, Relationship Types, or Compartment Types. Hence, the goal of this work is to develop a semi-automatic transformation from object-oriented and relational systems to role-based systems, based on the example of benchmarks. Benchmarks are comprised of a schema and data as well as query generators, parametrizable with respect to the number of supported types.
Research Goals
To achieve this goal the following subtasks must be carried out:
- Investigate different benchmarking techniques for information systems, e.g. TPC-C1, OLTP Benchmark2.
- Identify Natural Types, Role Types, Compartment Types, and Relationship Type in schema of the selected benchmark.
- Derive variants of role-based schemata from the benchmarks schema by iteratively including the these concepts.
- Based on the insights gained by the first three tasks, develop a semi-automatic transformation approach for schema, data and queries/transactions.
- Implement or derive a parametrizable benchmark including Schema Generator, Instance Generator, and Query Generator.
- Optionally, produce the schema as a CROM3 and its instances as CROI.
- Evaluate the benchmark by comparing its results with the selected Benchmark, measuring the query complexity with respect to the number of supported types, and conducting a performance analysis of the extended H2 DBMS4.
- Optionally, compare the semi-automatic separate of concerns to derive role-based systems w.r.t. aspect-oriented systems.
Footnotes
Betreuer: Thomas Kühn-:#-#:- Tobias Jäckel