|
EPSRC Network on Evolvability in Biology & Software SystemsSoftware Evolution and Evolutionary Computation Symposium Abstracts
University of Hertfordshire, Hatfield, U.K.
|
TIM FRENCH
Department of Computing and Information Systems, Park Square, Luton, Bedfordshire LU1 3JU, U.K.
SIMON POLOVINA
Department of Mathematical and Computing Sciences, Goldsmiths College, University of London, Lewisham Way, New Cross, London SE14 6NW, UK
The dynamic, turbulent environment in which Ecommerce systems operate epitomise that traditional system approaches to requirements elicitation, with its focus on the rigorous construction of a formal requirements catalogue then iterative refinement, are unable to deliver the goods to the system's users and its other heterogeneous stakeholder groups (Alexander, 1998; French and Polovina, 1999; Jarveenpaa and Tractinsky, 1999). We thus advocate an evolutionary systems development approach in which our starting requirements solution space (instantiated using object-oriented libraries of classes and objects) consists of a shared 'quanta' of meanings that are negotiated iteratively by all the system stakeholders until an initial 'agreement' is reached. We envisage that this process will necessitate the use of some form of semiotic agent (Rocha, 2000). These agents will assist human mediators in selecting appropriate solutions that match shared stakeholder knowledge, whilst acknowledging and modelling the situational nature of cultural and social forms of knowledge representation (using pre-defined collections of re-usable components). Such agents will generate and respond to user queries, whilst also separately identifying and modelling the particular situational contexts defined by users (i.e. social organisational and cultural factors), as well as forewarning redundant paths of information systems failure (Beynon-Davies, 1999).
Semiotic representations or discourses also fulfil our need to comfortably represent the ongoing, evolving look and feel of these systems (Liu, 2000). The quanta of shared meanings will, in effect, have formed a kind of genetic code or system blueprint (loosely comparable to genes in Biological systems) manifested as object components as their protein sequences. Thus the behavioural characteristics of the final software artefact will be a reflection of the optimal mix of these object library components. Unlike traditional object-orientation or biological DNA / RNA equivalents however, our quanta will contain embedded social and cultural semantics as an integral part of their definition. The eventual solution will be selected, component by component, so as to create a final 'organism' whose behavioural characteristics will already be at least be partially pre-optimised for a particular environment, whilst individual quanta will clearly be available for reuse and influence the initial starting point of the next project.
The mandate of our research therefore promotes what we coin as 'Shared Meanings Requirements Elicitation and Evolution' (SMREE), by which stakeholder groups negotiate dynamically shared-meanings using semiotic symbols and semantic tokens. Requirements engineering needs to take a bold leap forward and embrace ideas from both the Biology of evolutionary system and applied semiotics, in order to avoid an otherwise sterile closed world 'reductionist' future. Indeed, we are inclined to agree with Loomes and Jones (1999), that for too long, we have viewed requirements as being functional and static, rather than dynamic intangible entities defined by the various social and cultural forms of discourse which have created them. We seek therefore, to partially redress this balance somewhat, by borrowing, adapting and integrating ideas grounded in the field of the genetics of evolutionary systems (Rocha, 1998) with existing semiotic approaches to software engineering.