T. Yamaura’s research while affiliated with Hitachi, Ltd. and other places

This paper proposes new method to predict the number of the remaining bugs at the delivery inspection applied to every iteration of OOD, object oriented development. Our method consists of two parts. The first one estimates the number of the remaining bugs by applying the Gompertz curve. The second one uses the interval estimation called OOQP, object oriented quality probe. The basic idea of OOQP is to randomly extract a relatively small number of test cases, usually 10 to 20% of the entire test cases, and to execute them in the actual operation environment. From the test result of OOQP, we can efficiently predict the number of the remaining bugs by the interval estimation. The premier problem of OOQP is that OOD is imposed to use the system design specification document whose contents, like UML, tend to be ambiguous. Our estimation method works well at a matrix-typed organization where a QA team and a development team collaboratively work together to improve the software quality.

Due to business demands including cost cutting and schedule shortening, we see many software development projects that directly make software specifications without writing requirement specification. Such projects have many problems. This paper describes the ZC method which defines the requirement specification by visualizing the requirement features. In the ZC method, the objects to be developed are circled, and the relation among the objects are defined by lines: our empirical study discovered that a single A4 paper represents approximately 10 KLOC. We confirmed that the ZC method allows software engineers to easily and precisely define the user requirement. The ZC method is at the experimental stage, and we attempt it in the actual projects.

We raised issues of how the software education matching to a social request should be carried out in universities in this cyber-world era. In the high-tech era, it is unavoidable that almost all the high-tech products are provided as black-box. For this reason, it, too, is unavoidable that the education in high-tech era will, in stead of extracting the befit of the high-tech stuff, force us to memorize the "manners" which were institutionalized by a few high-tech gurus. As basic tools, everybody has to learn word processors, such as Word and Ichitaro, PowerPoint and Excel, and LATEX including tgif. On the other hand, special subjects, such as programming languages, programming language theory, compiler theory, OS theory, DBMS, Internet technology, business models, and software engineering are needed when becoming a software specialist. Mathematics, such as differentiation integral calculus, and physics, such as general dynamics must be also required as liberal arts of software science. The laws related to the patent, copyright, and accountancy for a floatation, and the logical thinking method are the subjects which should be studied as an application subject. In order to prepare the educational environment to study this wide scope of subjects, we must tackle the issues such as restriction of the number of credits and maintenance of lecturer staff. Such issues may be solved by applying less strict criteria when hiring lecturers, reexamining of the whole curriculum, or changing the system of company examination when employing new hires. This paper shows some proposals and remedies applicable to the universities in Japan.

One of the most important issues in the software development is how to guarantee that the software satisfies the quality defined in the requirement specification. This paper proposes that the issue can be solved, first the number of test cases is statistically calculated from the failure density defined in the requirement specification, then the selected test cases are executed basing on the hypothesis testing. This paper also presents how our method can be used for debugging. When the number of the test cases is calculated, we applied the statistical behavior of the software quality to the integration testing. We, however, did not consider the ripple effect since it is unable to measure. In order to guarantee the quality of 4σ, and 5σ, we found that many more test cases are needed than is previously believed enough.

Software quality is a serious issue for developers and customers. At Hitachi Software, we have used our extensive experience to improve quality assurance and significantly decrease our fault rates