On the applicability of problem frames to Web-based business applications

Problem frames offer a way to understand complex domains, hence their application to web based systems would seem a natural progression. However, it has been argued by some that web based systems are fundamentally different to typical software applications, and that existing problem frames may need extension to be applied to web-based applications. This paper discusses the application of problem frames to web-based business systems. We give an overview of web-based applications and introduce a classification of web business models. We then discuss some of the specific issues that have been suggested as points of differentiation between web based and traditional development. Finally, we suggest that problem frames are applicable to web based systems, but that it is the relative unfamiliarity of the domain which causes problems – ironically one reason for persevering with an approach which offers the potential for enhanced domain understanding. 1. Problem Frames and Web based Business Systems Web-based business systems are now important; not just for the large enterprise, but also for those small and medium enterprises, who find that a 'web presence' can be a competitive advantage. However, ‘the increased complexity of sites, in conjunction with the current web culture and ad hoc method usage, is creating a web crisis’ [1]. This has given rise to the term ‘tangled web’ implying that the web is a ‘morass’ of poorly developed, complex applications developed with a high degree of failure [2]. The importance of domain understanding has been accepted as a critical factor in obtaining complete and accurate requirements [3]. Webbased business systems are often unfamiliar, complex, and mission critical, and thus requirements activity is crucial. Hence, the use of problem frames, which offer a pathway through understanding and describing complex domains, would seem to be a natural step. However, there are suggestions that the set of existing problem frames may not cover all eventualities (e.g., Bray’s suggestion of the need for a simulator frame [4]). In particular it has been suggested that web-based business systems differ in fundamental ways, and that existing problems frames will be unable to be cater for development of web-based systems [5]. To consider this issue, we first discuss the nature of web-based applications, in terms of their underlying business models (section two). We then consider specific development differences (section 3), from the perspectives of users, developers and environment, before offering some brief conclusions (section 4). 2. Web based Business Models and Categories To produce a successful web application it is essential to understand the problem domain. However, understanding the (wider) e-business domain requires an appreciation of business models and categorisations that are not yet stable [6], and are still the subject of considerable discussion in the literature [7]. These business models are important because they enable 1 E-business is considered as business using Internet technologies, whereas web-based applications specifically utilise web technology to this end. developers to understand and apply the different delivery channels and infrastructure that may be required. Rappa [7] believes that a business model should specify where a business lies in the value chain, but considers that business models are the most discussed and least understood aspect of web development methods. For example, Business to Consumer (B to C) is a frequently used term, which often means different things to different people. Drew [8] believes that most of the literature discusses the concept of a business model in terms of a customer value proposition, pricing strategy or a particular Internet technology and that a holistic view has yet to emerge. Furthermore, many categorisations can be misleading in that they focus on what amounts to the presentation of the application, rather than the business process supported. However, some consensus is starting to emerge. For example, Weill and Vitale [9] have produced a useful categorisation of atomic ebusiness models (Table 1). Table 1: Categories of web application (taken from [10]) This approach recognises that business context is important to understand the nature of a web-based system, and considers which of 70 infrastructure services are necessary to conduct business electronically. These 70 services are divided into 9 categories: Applications Infrastructure (1) Communication (2), Data Management (3), IT Management (4), Security (5), Architecture (6), Channel Management (7), IT Research (8) and IT Education (9), across eight fundamental business models. This categorisation appears to offer a (relatively) exhaustive survey of web-application types. A complete analysis of which frames would apply to each of these categories is beyond the scope of this paper. However, our work so far suggests that problem frames are no more difficult to apply here then they are in many other contexts. The apparent difficulty, we suggest, being related to the immaturity of the domain and the lack of a common framework (and agreed vocabulary) for business models. Hence, since consideration of the breadth of web-application types does not (in itself) suggest fundamental differences, we now consider differences suggested within the literature. That is, arguments that the process of developing web-applications is fundamentally different from traditional software engineering. 3. Web applications versus traditional applications A number of authors have suggested that there are fundamental differences between web development and software development, and that these differences may require additional problem frames. For example, Cox and Phalp [5] outline a scenario detailing a customer application made over the Internet. They argue that one aspect of their scenario; a credit checking transaction, does not fit with existing frames. However, the argument relies on the use of a customer domain, and the authors concede this as a possible misuse of the problem frames approach. Furthermore, it could be argued that this particular transaction type is one that would be encountered within a typical 'traditional' application, that it is not unique to web-based applications Of course, this is just a single example, but it illustrates that the argument for the applicability of problem frames hinges upon whether web-based business systems are fundamentally different or just another example of a complex software engineering domain. Clearly, it is not possible to specifically address every argument that we have encountered within the literature. However, there appear to be some common themes. Hence, we have categorised arguments into those relating to: user domain, developer, and environment. Brief discussions of these arguments are given in the following sections. (Note that for reasons of space, we have chosen illustrative examples for each category). E-business model 1 2 3 4 5 6 7 8 9 Content Provider ++ ++ ++ * + +