Neil A. M. Maiden

Requirements engineering paper classification and evaluation criteria: a proposal and a discussion

In recent years, members of the steering committee of the IEEE Requirements Engineering (RE) Conference have discussed paper classification and evaluation criteria for RE papers. The immediate trigger for this discussion was our concern about differences in opinion that sometimes arise in program committees about the criteria to be used in evaluating papers. If program committee members do not all use the same criteria, or if they use criteria different from those used by authors, then papers might be rejected or accepted for the wrong reasons. Surely not all papers should be evaluated according to the same criteria. Some papers describe new techniques but do not report on empirical research; others describe new conceptual frameworks for investigating certain RE problems; others report on industrial experience with existing RE techniques. Other kinds of papers can also be easily recognized. All of these types of papers should be evaluated according to different criteria. But we are far from a consensus about what classes of paper we should distinguish, and what the criteria are for each of these classes.

Supporting scenario-based requirements engineering

Scenarios have been advocated as a means of improving requirements engineering yet few methods or tools exist to support scenario based RE. The paper reports a method and software assistant tool for scenario based RE that integrates with use case approaches to object oriented development. The method and operation of the tool are illustrated with a financial system case study. Scenarios are used to represent paths of possible behavior through a use case, and these are investigated to elaborate requirements. The method commences by acquisition and modeling of a use case. The use case is then compared with a library of abstract models that represent different application classes. Each model is associated with a set of generic requirements for its class, hence, by identifying the class(es) to which the use case belongs, generic requirements can be reused. Scenario paths are automatically generated from use cases, then exception types are applied to normal event sequences to suggest possible abnormal events resulting from human error. Generic requirements are also attached to exceptions to suggest possible ways of dealing with human error and other types of system failure. Scenarios are validated by rule based frames which detect problematic event patterns. The tool suggests appropriate generic requirements to deal with the problems encountered. The paper concludes with a review of related work and a discussion of the prospects for scenario based RE methods and tools.

Acquiring COTS software selection requirements

Commercial off the shelf software can save development time and money if you can find a package that meets your customers needs. The authors propose a model for matching COTS product features with user requirements. To support requirements acquisition for selecting commercial off the shelf products, we propose a method we used recently for selecting a complex COTS software system that had to comply with over 130 customer requirements. The lessons we learned from that experience refined our design of PORE (procurement oriented requirements engineering), a template based method for requirements acquisition. We report 11 of these lessons, with particular focus on the typical problems that arose and solutions to avoid them in the future. These solutions, we believe, extend state of the art requirements acquisition techniques to the component based software engineering process.

ACRE: selecting methods for requirements acquisition

A framework is presented that assists requirements engineers in choosing methods for requirements acquisition. Practitioners are often unaware of the range of methods available. Even when practitioners are aware, most do not foresee the need to use several methods to acquire complete and accurate requirements. One reason for this is the lack of guidelines for method selection. The ACRE framework sets out to overcome these limitations. Method selection is achieved using questions driven from a set of facets which define the strengths and weaknesses of each method. The framework provides guidelines for requirements engineering practitioners. It has undergone evaluation through its presentation to highly experienced requirements engineers. Some results from this evaluation have been incorporated into the presented version of ACRE.

A proposal for a scenario classification framework

The requirements engineering, information systems and software engineering communities recently advocated scenario-based approaches which emphasise the user/system interaction perspective in developing computer systems. Use of examples, scenes, narrative descriptions of contexts, mock-ups and prototypes-all these ideas can be called scenario-based approaches, although exact definitions are not easy beyond stating that these approaches emphasise some description of the real world. Experience seems to tell us that people react to ‘real things’ and that this helps in clarifying requirements. Indeed, the widespread acceptance of prototyping in system development points to the effectiveness of scenario-based approaches. However, we have little understanding about how scenarios should be constructed, little hard evidence about their effectiveness and even less idea about why they work.The paper is an attempt to explore some of the issues underlying scenario-based approaches in requirements engineering and to propose a framework for their classification. The framework is a four-dimensional framework which advocates that a scenario-based approach can be well defined by itsform, content, purpose andlife cycle. Every dimension is itself multifaceted and a metric is associated with each facet. Motivations for developing the framework are threefold: (a) to help in understanding and clarifying existing scenario-based approaches; (b) to situate the industrial practice of scenarios; and (c) to assist researchers develop more innovative scenario-based approaches.

Exploiting reusable specifications through analogy

domain model representing critical determinants of the theater reservation/course administration analogy Objects in Objects in space space Monitor objects whenever one moves Cancelled allocations System should monitor object spaces to ensure only one object in any space Abstract domain model representing critical determinants of the air traffic control/flexible manufacturing analogydomain model representing critical determinants of the air traffic control/flexible manufacturing analogy FigUre 3. Abstract domain models representing critical determinants of the two example analogies. frequent mistake made by the relatively inexperienced software engineers in this study was to focus on surface, lexical properties of the reusable specification, whereas successful reuse requires comprehension of deeper analogous concepts. Analogies represented in abstract form proved easier to recognize, although once recognized, reuse of concrete specifications proved more effective. Reuse of concrete specifications supported by abstraction during analogical recognition is one scenario which may lead to effective specification

The implementation of enterprise resource planning packages in different organisational and national cultures

Abstract Enterprise resource planning (ERP) packages provide generic off-the-shelf business and software solutions to customers. However, these packages are implemented in companies with different organisational and national cultures, and there is growing evidence that failure to adapt ERP packages to fit these cultures leads to projects that are expensive and late. This paper describes research that synthesises social science theories of culture to handle the impact of culture on ERP package implementation more efficiently. It describes a knowledge meta-schema for modelling the surface and the deeper manifestations of culture. It reports an empirical study into the implementation of SAP R/3s sales and distribution (SD) module in a large pharmaceuticals organisation in Scandinavia and the UK. Results provide evidence for an association between organisational culture and ERP implementation problems but no direct evidence for an association between national culture and implementation problems. Furthermore, results demonstrate that these diverse implementation problems can be caused by a mismatch between a small set of core values indicative of a customers organisational culture. At the end of the paper, our predictions are reviewed, conclusions are made about them and about the work of the key authors of national and organisational culture, and future work is discussed.

Towards a framework for integrating agile development and user-centred design

Due to a number of similarities between user-centred design (UCD) and agile development, coupled with an appreciation that developers are rarely usability experts, it seems attractive to integrate these two approaches. However, although agile methods share some of the same aims as UCD, there are also distinct differences. These differences have made the use of these methods on development projects problematic. This paper reports a field study designed to investigate the use of agile methods alongside UCD in one particular organization. The aim of the study was to develop a framework for use by project teams wishing to integrate UCD practices with agile development. The study, its findings and five principles for integrating UCD and agile development arising from this work are discussed.

Implementing enterprise resource planning packages in different corporate and national cultures

ERP (enterprise resource planning) packages provide generic off-the-shelf business and software solutions for customers. However, these packages are implemented in companies with different corporate and national cultures and there is growing evidence that failure to adapt ERP packages to fit these cultures leads to projects which are expensive and overdue. This paper describes research which synthesizes social science theories of culture in order to be able to model and predict the impact of culture on ERP package implementation. It describes a knowledge meta-schema for modelling the surface and deeper manifestations of culture and predictions of ERP implementation problems based on national culture differences. It reports on an empirical study into the implementation of ERP packages in a large pharmaceuticals organization in Scandinavia and the UK. The results provide evidence for an association between corporate culture and ERP implementation problems but no direct evidence for an association between national culture and implementation problems. Furthermore, the results demonstrate that these diverse implementation problems can be caused by a mismatch between a small set of core values which are indicative of a customers corporate culture. The paper concludes with revisions to the design of our extended method for ERP package implementation to the design of the meta-model and to rules that codify culture constraints that are applied in order to analyse instances of the meta-model.

CREWS-SAVRE: Scenarios for Acquiring and Validating Requirements

This paper reports research into semi-automatic generation of scenarios for validating software-intensive system requirements. The research was undertaken as part of the ESPRIT IV 21903 ‘CREWS’ long-term research project. The paper presents the underlying theoretical models of domain knowledge, computational mechanisms and user-driven dialogues needed for scenario generation. It describes how CREWS draws on theoretical results from the ESPRIT III 6353 ‘NATURE’ basic research action, that is object system models which are abstractions of the fundamental features of different categories of problem domain. CREWS uses these models to generate normal course scenarios, then draws on theoretical and empirical research from cognitive science, human-computer interaction, collaborative systems and software engineering to generate alternative courses for these scenarios. The paper describes a computational mechanism for deriving use cases from object system models, simple rules to link actions in a use case, taxonomies of classes of exceptions which give rise to alternative courses in scenarios, and a computational mechanism for generation of multiple scenarios from a use case specification.