Paul Feyerabend und die Softwaretechnologie

Zusammenfassung   Der folgende Beitrag ist ein Pl?doyer für st?rkere methodische Standards in der Softwaretechnologie. Nicht immer werden wissenschaftstheoretischen Grunds?tze beachtet wie das Prinzip, da? Aussagen falsifizierbar sein sollten. Dies scheint die These der sog. Konstruktivisten zu best?tigen, da? n?mlich „objektive Wahrheiten“ in Wirklichkeit soziale Konstrukte seien. Der Beitrag pl?diert deshalb für eine st?rkere empirische Fundierung der Softwaretechnologie. Eingegangen in überarbeiteter Form am 07.09.1998     

Training trajectories by continuous recurrent multilayer networks

This paper addresses the problem of training trajectories by means of continuous recurrent neural networks whose feedforward parts are multilayer perceptrons. Such networks can approximate a general nonlinear dynamic system with arbitrary accuracy. The learning process is transformed into an optimal control framework where the weights are the controls to be determined. A training algorithm based upon a variational formulation of Pontryagin's maximum principle is proposed for such networks. Computer examples demonstrating the efficiency of the given approach are also presented.     

Generic Polynomials with Few Parameters

We call a polynomial g(t₁, . . . , t_m,X ) over a field K generic for a group G if it has Galois group G as a polynomial inX , and if every Galois field extension N / L withK L and Gal(N / L) ≤ G arises as the splitting field of a suitable specializationg (λ₁, . . . , λ_m, X) withλ_i L. We discuss how the rationality of the invariant field of a faithful linear representation leads to a generic polynomial which is often particularly simple and therefore useful. Then we consider various examples and applications in characteristic 0 and in positive characteristic. These include results on so-called vectorial polynomials and a generalization of an embedding criterion given by Abhyankar. We give recursive formulas for generic polynomials over a field of defining characteristic for the groups of upper unipotent and upper triangular matrices, and explicit formulae for generic polynomials for the groups GU₂(q²) andGO₃ (q).     

Protecting shared information in networks: A network security game with strategic attacks

A digital security breach, by which confidential information is leaked, does not only affect the agent whose system is infiltrated but is also detrimental to other agents socially connected to the infiltrated system. Although it has been argued that these externalities create incentives to underinvest in security, this presumption is challenged by the possibility of strategic adversaries that attack the least protected agents. In this paper we study a new model of security games in which agents share tokens of sensitive information in a network of contacts. The agents have the opportunity to invest in security to protect against an attack that can be either strategically or randomly targeted. We show that, in the presence of random attack, underinvestments always prevail at the Nash equilibrium in comparison with the social optimum. Instead, when the attack is strategic, either underinvestments or overinvestments are possible, depending on the network topology and on the characteristics of the process of the spreading of information. Actually, agents invest more in security than socially optimal when dependencies among agents are low (which can happen because the information network is sparsely connected or because the probability that information tokens are shared is small). These overinvestments pass on to underinvestments when information sharing is more likely (and therefore, when the risk brought by the attack is higher). In order to keep our analysis tractable, some of our results on strategic attacks make an assumption of homogeneity in the network, namely, that the network is vertex‐transitive. We complement these results with an analysis on star graphs (which are nonhomogeneous), which confirms that the essential lines of our findings can remain valid on general networks.     

A multi-objective approach to the application of real-world production scheduling

An initiative was introduced in one of the production facilities of Germany’s E.G.O. Group in order to enhance its SAP information system with a custom-made application for production-scheduling optimization. The goal of the optimization is to find a production schedule that satisfies different, contradictory production and business constraints. We show the challenges faced in the application of the multi-objective optimization approach, which is gaining influence in the management of production scheduling. We implement a memetic version of the Indicator-Based Evolutionary Algorithm with customized reproduction operators and local search procedures to find a set of feasible, non-dominated solutions. Such a memetic algorithm was applied to two real order lists from the production company. Additionally, we also lay out an efficient presentation of the multi-objective results for an expert’s support in decision making. This provides the management with the possibility to gain additional insights into how the production schedule dynamically reacts to changes in the decision criteria. We show that the multi-objective approach is able to find high-quality solutions, which enables flexibility when it comes to quickly adapting to specific business conditions.     

Development of a computer-based interviewing tool to enhance the requirements gathering process

Software design is an area where it is most important to elicit an accurate and complete set of requirements. Determining the users requirements and their qualitative satisfaction with a developing product are two major problems which arise, often due to the different backgrounds and perspectives of the different stakeholders, including the software engineer, the domain expert and the end-user. Involving users throughout the design and development process is an essential part of any project. Computer-based interviewing has been found to be a valuable tool in eliciting information, which can also enhance interpersonal communication afterwards. This paper discusses the development of a computer interviewing tool to facilitate the gathering of user requirements and conducting user evaluations.     

User-Sensitive Inclusive Design

Although “User-Centred”, “Participatory”, and other similar design approaches have proved to be very valuable for mainstream design, their principles are more difficult to apply successfully when the user group contains, or is composed of, older and/or disabled users. In the field of design for older and disabled people, the “Universal Design”, “Inclusive Design” and “Design for All” movements have encouraged designers to extend their design briefs to include older and disabled people. The downside of these approaches is that they can tend to encourage designers to follow a traditional design path to produce a prototype design, and only then investigate how to modify their interfaces and systems to cope with older and/or disabled users. This can lead to an inefficient design process and sometimes an inappropriate design, which may be “accessible” to people with disabilities, but in practice unusable. This paper reviews the concept that the authors have called “User-Sensitive Inclusive Design”, which suggests a different approach to designing for marginalised groups of people. Rather than suggesting that designers rely on standards and guidelines, it is suggested that designers need to develop a real empathy with their user groups. A number of ways to achieve this are recommended, including the use of ethnography and techniques derived from professional theatre both for requirements gathering and for improving designers’ empathy for marginalised groups of users, such as older and disabled people.     

TORC: test plan optimization by requirements clustering

Acceptance testing is a time-consuming task for complex software systems that have to fulfill a large number of requirements. To reduce this effort, we have developed a widely automated method for deriving test plans from requirements that are expressed in natural language. It consists of three stages: annotation, clustering, and test plan specification. The general idea is to exploit redundancies and implicit relationships in requirements specifications. Multi-viewpoint techniques based on RM-ODP (Reference Model for Open Distributed Processing) are employed for specifying the requirements. We then use linguistic analysis techniques, requirements clustering algorithms, and pattern-based requirements collection to reduce the total effort of testing against the requirements specification. In particular, we use linguistic analysis for extracting and annotating the actor, process and object of a requirements statement. During clustering, a similarity function is computed as a measure for the overlap of requirements. In the test plan specification stage, our approach provides capabilities for semi-automatically deriving test plans and acceptance criteria from the clustered informal textual requirements. Two patterns are applied to compute a suitable order of test activities. The generated test plans consist of a sequence of test steps and asserts that are executed or checked in the given order. We also present the supporting prototype tool TORC, which is available open source. For the evaluation of the approach, we have conducted a case study in the field of acceptance testing of a national electronic identification system. In summary, we report on lessons learned how linguistic analysis and clustering techniques can help testers in understanding the relations between requirements and for improving test planning.