AbstractsComputer Science

Data processing software is an essential component of systems that aggregate and analyse real-world data, thereby enabling automated interaction between such systems and the real world. In data processing systems, inputs are often big and complex files that have a well-defined structure, and that often have dependencies between several of their fields. Testing of data processing systems is complex. Software engineers, in charge of testing these systems, have to handcraft complex data files of nontrivial size, while ensuring compliance with the multiple constraints to prevent the generation of trivially invalid inputs. In addition, assessing test results often means analysing complex output and log data. Complex inputs pose a challenge for the adoption of automated test data generation techniques; the adopted techniques should be able to deal with the generation of a nontrivial number of data items having complex nested structures while preserving the constraints between data fields. An additional challenge regards the automated validation of execution results. To address the challenges of testing data processing systems, this dissertation presents a set of approaches based on data modelling and data mutation to automate testing. We propose a modelling methodology that captures the input and output data and the dependencies between them by using Unified Modeling Language (UML) class diagrams and constraints expressed in the Object Constraint Language (OCL). The UML class diagram captures the structure of the data, while the OCL constraints formally describe the interactions and associations between the data fields within the different subcomponents. The work of this dissertation was motived by the testing needs of an industrial satellite Data Acquisition (DAQ) system; this system is the subject of the empirical studies used within this dissertation to demonstrate the application and suitability of the approaches that we propose. We present four model-driven approaches that address the challenges of automatically testing data processing systems. These approaches are supported by the data models generated according to our modelling methodology. The results of an empirical evaluation show that the application of the modelling methodology is scalable as the size of the model and constraints was manageable for the subject system. The first approach is a technique for the automated validation of test inputs and oracles; an empirical evaluation shows that the approach is scalable as the input and oracle validation process executed within reasonable times on real input files. The second approach is a model-based technique that automatically generates faulty test inputs for the purpose of robustness testing, by relying upon generic mutation operators that alter data collected in the field; an empirical evaluation shows that our automated approach achieves slightly better instruction coverage than the manual testing taking place in practice. The third approach is an evolutionary algorithm to automate the… Advisors/Committee Members: Briand, Lionel [superviser].