Web applications are increasingly developed in many domains and most were implemented by the ad hoc approach. However, various quality metrics become progressively more difficult to measure and mange as the developing application size and scope grows. Among the typical non-functional characteristics, in this research, factors of link validity will be deeply investigated. Essentially link validity includes various characteristics such as correctness, relevance, completeness and integrity implicitly, which will deeply affect the effectiveness and efficiency on information retrieval and browsing.
In most web applications, it is unlikely that the application developers may validate all possible navigation paths. In this paper, a framework of quantitative certification on link validity will be proposed to ensure that all linked paths provide consistent and reasonable information streams and appropriate contexts.
In this research, the rational for statistical usage testing is investigated and employed to certify links for a website. In principle, a software usage model is first established, which characterizes various operational uses of a software system. Suppose that an operational use is a skeleton for the intended use of the software in an intended environment. Thus, all possible operational uses of a software system will constitute a population with a huge size. If a usage sample of test cases is drawn statistically from the usage population, performance on this sample may then be used as a basis for the evaluation of software quality.
Under the web-browsing environments, all possible navigation paths are first formulated to represent a usage model with the Markov chain property, that is then analyzed and used to generate test script file statistically. The proposed mechanism is not only systematic on certifying effectively hyper links, but also efficient highly for those complex information structures. A real application of the proposed approach to a web application will also be demonstrated quantitatively through a certification tool ToolCertify.
Associate Professor at Software Engineering Laboratory in Tunghai University
Points of interest: software engineering, software quality assurance, software quality certification, project management