asp.net综合类B2B市场的设计与实现 第5页C#英文文献
reliability, most of available software reliability techniques fail to deliver their promises． Significantly better software reliability techniques are required． Second，besides being closely related to hardware reliability researches，the field of software reliability should be interpreted in a wide sense to cover disparate techniques in various research areas such as concurrent computing，model checking，real-time scheduling，intelligent decision—making，and so on．This is reflected in diverse topics covered in annual international symposium on software reliability engineering．
In 1990s, safety—critical software became a major theme in software reliability researches． How to guarantee the reliability of safety—critical software is a grand challenge to human intelligence． Since then，new kinds of applications including component—based software and Web software emerge as research topics for software reliability．However it should be noted that among the various research topics，the mainstream techniques and applications can no longer be identified easily.
Existing researches on software reliability can be divided into two parts：model—based studies and empirical studies．
www.751com.cn subsection reviews model—based studies，whereas empirical studies are left to the next subsection．Model-based studies can be treated as synonymous with theory inventing to a large extent．However we stick to the former terminology since the notion of models is extensively adopted in software reliability researches．
A model—based study begins with making a set of assumptions for software reliability．Roughly speaking，it comprises the following steps：
1 A set of assumptions is made to abstract the software process or system under study to a certain level；ideally, the assumptions should fit engineering practice and be mathematically tractable．
2 A mathematical model is obtained, which can be analytic，stochastic，or formal．
3The mathematical model is investigated to reveal the properties of interest and／or to derive measures of interest in a mathematically rigorous manner；consequently，theoretical results are obtained．
4 The theoretical results are tested or validated against some form s of evidence including observed data collected from real software projects or simulation settings．
The underlying philosophy of model—based studies is that a mathematically tractable model exists for the problem under investigation．Model is the central object of concern, and the mathematical analysis of the properties of the obtained model plays a key role.Besides various software reliability models and others．model—based testing and software model checking also fall into the scope of model—based studies．The advantages of model—based studies lie in that the vast and rich library of existing mathematical tools can be exploited and the mathematical analysis of the given model m ay generate useful hints for developing m ore powerful models。Remember that in—depth researches for software reliability problems should eventually be based on or lead to appropriate mathematical models． However the disadvantages of model—based studies are also obvious．The underlying mathematical assumptions are not easy to be validated in engineering practice which is often much more complicated than what a mathematically tractable model may describe．The derived mathematical properties may or may not be useful for software process improvements.
Although empirical studies are reported for software reliability from time to time in scattered journals and conference proceedings, they are not the majority of software reliability researches. This contrasts with the growing awareness and widely accepted importance of empirical studies in software engineering in general ,which are motivated by the fact that more often than not , various proposals for software process improvements and system optimizations were made without being backed by hard evidence or objective evaluation techniques for their effectiveness or advantages . An empirical study is an act or operation on practical software processes or artifacts, in which data of interest are generated, collected, analyzed, and interpreted with the purpose of discovering, testing , customerizing, and improving certain known or unknown theories
In this broad sense empirical studies cover experimental researches as well as engineering researches． They can be observational，historical，and controlled.
Unlike model—based studies which are centered around mathematical models, an empirical study focuses on practical software processes and／or systems and the corresponding data of interest．The underlying philosophy is that theories should begin and end with practical subjects and do not make sense in isolation．A general procedure for an empirical study is as follows．
(1) Select factors or theories whose effects are of interest to be examined．
(2)Select subject software processes or experiments on which the effects of the factors or theories are to be examined．
(3) Build up the practical scenarios or experimental platforms from which data of interest are to be generated and／or collected．
(4) Analyze the collected data to discover possible theories or examine hypothesized theories．
In a recent empirical study of regression testing，four factors (Test Suite Granularity，Test Input Grouping, Regression Testing Technique, Interactions)were considered and four hypotheses were examined as follows．
H1(Test Suite Granularity)：Test suite granularity does not have a significant impact on the costs and benefits of regression testing techniques．
H2(Test Input Grouping)：Test input grouping does not have a significant impact on the costs and benefits of regression testing techniques．
H3 (Technique)：Regression testing techniques do not perform significantly differently in terms of the selected costs and benefits measures．
H4(Interactions)：Test suite granularity and test input grouping effects across regression testing techniques and programs do not significantly differ．
Four regression testing techniques．namely retest—al1, regression test selection，test suite reduction，and test case prioritization were then applied to
two subject programs：Emp—Server and Bash with each comprising nine consecutive versions．The testing data were collected and analyzed to examine the four hypotheses．
The main advantage of empirical studies is that theories of concern can be validated or invalidated by hard evidence with respect to practical software processes and/or products if the empirical studies are repeatable and realistic． This renders valid theories to be useful for software process improvements and system optimization． However．theories m ay be mis—validated or mis--invalidated if the collected data are not repeatable or realistic in some sense, and thus may deliver misleading guidelines for software process improvements and system optimization． This is sensitive since a large number of factors may play a role in an empirical study．It is not easy to distinguish important factors from unimportant factors．Further, human are ofen involved in data generation processes and the repeatability of the collected data is at risk． Empirical studies can validate or invalidate theories，but cannot prove theories．The disadvantages of empirical studies are obvious． An additional disadvantage of empirical studies is that they may incur high costs and are time--consuming in building experimental platforms and generating/collecting data of interest.
SOFTWARE MAINTENANCE
Maintenance activities involve making enhancements[1] to software products，adapting products to new environments，and correcting problems．Software product enhancement may involve providing new functional capabilities，improving user displays and modes of interaction，upgrading external documents and internal documentation，or upgrading the performance characteristics of a system．上一页 [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] ... 下一页 >>
asp.net综合类B2B市场的设计与实现 第5页下载如图片无法显示或论文不完整，请联系qq752018766
