Mario Jino

AUTOMATIC TEST DATA GENERATION FOR PROGRAM PATHS USING GENETIC ALGORITHMS

A new technique and tool are presented for test data generation for path testing. They are based on the dynamic technique and on a Genetic Algorithm, which evolves a population of input data towards reaching and solving the predicates along the program paths. We improve the performance of test data generation by using past input data to compose the initial population for the search. An experiment was done to assess the performance of the techniques compared to that of random data generation.

Identification of potentially infeasible program paths by monitoring the search for test data

A tool and techniques are presented for test data generation and identification of a paths likely unfeasibility in structural software testing. The tool is based on the dynamic technique and search using genetic algorithms. Our work introduces a new fitness function that combines control and data flow dynamic information to improve the process of search for test data. The unfeasibility issue is addressed by monitoring the genetic algorithms search progress. An experiment shows the validity of the developed solutions and the benefit of using the tool.

Diversity oriented test data generation using metaheuristic search techniques

We present a new test data generation technique which uses the concept of diversity of test sets as a basis for the diversity oriented test data generation - DOTG. Using DOTG we translate into an automatic test data generation technique the intuitive belief that increasing the variety, or diversity, of the test data used to test a program can lead to an improvement on the completeness, or quality, of the testing performed. We define the input domain perspective for diversity (DOTG-ID), which considers the distances among the test data in the program input domain to compute a diversity value for test sets. We describe metaheuristics which can be used to automate the generation of test sets for the DOTG-ID testing technique: simulated annealing; a genetic algorithm; and a proposed metaheuristic named simulated repulsion. The effectiveness of DOTG-ID was evaluated by using a Monte Carlo simulation, and also by applying the technique to test simple programs and measuring the data-flow coverage and mutation scores achieved. The standard random testing technique was used as a baseline for these evaluations. Results provide an understanding of the potential gains in terms of testing effectiveness of DOTG-ID over random testing and also reveal testing factors which can make DOTG-ID less effective.

Extracting Information from Experimental Software Engineering Papers

Experiments have been conducted to investigate analysis, design, implementation, testing, maintenance, quality assurance and reuse techniques, but, a body of evidence has not yet been built that enables a project manager to know with confidence what software processes produce what product characteristics and under what conditions. This paper extends an approach we proposed earlier to extract information from papers so that systematically analyzing results from several papers is possible. It also describes an in-vitro experiment we did with graduate students to validate the approach. The results show that the approach is feasible and can be taught to less experienced researchers.

Improving random test sets using the diversity oriented test data generation

We present a measure that characterizes the diversity of a test set from the perspective of the input domain of the program under test. By using a metaheuristic algorithm, randomly generated test sets (RTS) are evolved towards Diversity Oriented Test Sets (DOTS), which thoroughly cover the input domain. DOTS are evaluated using a Monte Carlo simulation to assess how testing factors influence their effectiveness and also by the values of data flow coverage and mutation scores attained on simple programs. Results provide understanding on possible gains of using DOTS and on circumstances where RTS can be more effective.

Automated Information Extraction from Empirical Software Engineering Literature: Is that possible?

The number of scientific publications is constantly increasing, and the results published on empirical software engineering are growing even faster. Some software engineering publishers have begun to collaborate with research groups to make available repositories of software engineering empirical data. However, these initiatives are limited due to data ownership and privacy issues. As a result, many researchers in the area have adopted systematic reviews as a mean to extract empirical evidence from published material. Systematic reviews are labor intensive and costly. In this paper, we argue that the use of information extraction tools can support systematic reviews and significantly speed up the creation of repositories of SE empirical evidence.

A testing approach for XML schemas

XML is a language frequently used for data representation and interchange in Web-based applications. In most cases, the XML documents must conform to a schema that defines the type of data that is accepted by a Web application. In this sense, an error in the schema or in the XML document can lead to failures in the application; the use of testing approaches, criteria and specific tools to ensure the reliability of data in the XML format is fundamental. We present a testing approach that helps to reveal faults in XML schemas. The test process involves generating XML documents with some modifications with respect to the original XML document and using queries to these documents to validate the schema. The XML documents and queries are generated according to a set of fault classes defined for the XML schemas. A case study applying the proposed approach is described and the results are presented.

Using Context Distance Measurement to Analyze Results across Studies

Providing robust decision support for software engineering (SE) requires the collection of data across multiple contexts so that one can begin to elicit the context variables that can influence the results of applying a technology. However, the task of comparing contexts is complex due to the large number of variables involved. This works extends a previous one in which we proposed a practical and rigorous process for identifying evidence and context information from SE papers. The current work proposes a specific template to collect context information from SE papers and an interactive approach to compare context information about these studies. It uses visualization and clustering algorithms to help the exploration of similarities and differences among empirical studies. This paper presents this approach and a feasibility study in which the approach is applied to cluster a set of papers that were independently grouped by experts.

Applying Extended Finite State Machines in Software Testing of Interactive Systems

Model Based Testing (MBT) is a functional testing technique that makes use of information from behavioral models of the software to carry out the testing task. This technique has been commonly used in testing of interactive systems, where the used model represents the system behavior reacting to user’s actions. Finite State Machines (FSMs) are one of the most used modeling techniques for MBT. However, traditional FSMs do not provide mechanisms to model important behavioral aspects of the software such as its data flow. This paper proposes an extension to the traditional FSMs, which provides data flow modeling mechanisms and is used as a basis to define a set of functional testing criteria, extending known structural testing criteria. Moreover, the application of the defined functional testing criteria is compared, through a practical experiment, to the application of their corresponding structural testing criteria – both applied as adequacy criteria.

Machine Learning Methods and Asymmetric Cost Function to Estimate Execution Effort of Software Testing

Planning and scheduling of testing activities play an important role for any independent test team that performs tests for different software systems, developed by different development teams. This work studies the application of machine learning tools and variable selection tools to solve the problem of estimating the execution effort of functional tests. An analysis of the test execution process is developed and experiments are performed on two real databases. The main contributions of this paper are the approach of selecting the significant variables for database synthesis and the use of an artificial neural network trained with an asymmetric cost function.