|
|
Semiotic Enterprise Design for IT Applications (SEDITA)* Motivation – improve organisational design for IT applications However technically excellent computer systems may be, they often add too little value to a business. Businesses tend to be coy about admitting these deficiencies but the evidence cannot be ignored. Individual cases, especially in the public sector, are reported every week: for example, Cabinet Office (2000) and the Parliamentary Select Committee on Public Accounts (2000). Macro-economic studies of productivity changes related to the use IT also reveal that the failure is general. Strassman (1985, 1990 and 1997) continues to demonstrate this in large samples of manufacturing firms and the US Department of Commerce report, The Digital Economy, (2000) shows that the massive benefits from selling IT products do not bring equally significant benefits for either the manufacturing or service sectors. A huge economic advantage is awaiting us if we can apply IT with improved organisational effectiveness. To address this problem, we propose to ensure a better match between the business and its IT systems. We shall do this by treating the business as an information system in a new methodology for the upstream phases of software engineering. This will tackle one root cause of the mismatch between business and IT: the lack of adequate tools for the job. Our thesis is that, in order to place the organisation first, it must be specified as an information system, regardless of any technology, but in a precise and formal way. Information requirements expressed in this formal business specification will make it easier to pull the technology into a seamless relationship with the organisation. This goal would be over-ambitious if we were starting from scratch but that is not so. Theoretical work began in the early 1970s with UK Research Council support, to understand how to describe organisations with formal precision, as information systems. This and other work has evolved into a new, internationally recognised discipline of Organisational Semiotics - hence the title of this proposal. (Semiotics is the study of signs. The notion of a sign is more fundamental than that of information, which is revealed as a vague concept with numerous precise meanings, which can be defined as properties of signs.) From the underlying theory, the MEASUR methodology has been evolving over the last 10 to 15 years (Stamper et al. 1988; Stamper 1994; Liu, 2000). The research on MEASUR, for a variety of reasons, had to be carried out in a fragmentary manner. At present we have all the essential techniques, which have been individually tested on real problems, but the work is not yet complete. (The MEASUR techniques have been partially tested in a diversity of ad hoc industrial situations, spread over firms in many countries. They include many SMEs, but also public sector bodies ranging from the Dutch Ministry of Internal Affairs to the Burkina Faso rural savings bank, and large organisations such as Mars Foods, Indiana National Bank, Eli Lilly, Philips, Heineken, Bajaj Auto, Electrolux, CREST UK Stock Settlements. ) The motivation then, combining industrial needs with current research results, is to complete the work. To succeed, we must bring the MEASUR method into mainstream use where IT is being applied to business problems. This means that we must a) integrate the techniques we have, adjusting them where necessary, b) validate the methodology as a whole on real industrial problems of sufficient size and complexity, c) provide supporting tools so that d) it will be acceptable to business users with e) a natural interface to the downstream software engineering phases. The proposed approach – building on prior results MEASUR captures the kernel of a business’s information requirements (the relevant organisational knowledge, concepts, terminology, meanings, responsibilities and priorities) with formality and precision so that the IT requirements (programs, reports, interfaces, performance, etc.) may be deduced logically. Eventually (but not a part of this project) we see that, at least, a default implementation could be generated automatically. Current methodologies treat business information systems as networks of data processes and information flows, without ever specifying why the information flows or what it means. MEASUR answers those questions. It does so within a new paradigm: instead of looking at information flows, it deals with information fields. An information field comprises the shared social norms (such as rules, regulations, and social and business conventions and the background culture) that act as a 'field of force' binding the norm-subjects into a pattern of organised behaviour. The business requirements are that these norms be followed but to apply them correctly, people need information, in the shape of any messages that provide the correct meanings. The meanings are actually dictated by the norms, the perceptual norms being the most basic. Current requirements analysis methods only treat meanings informally and move directly to studying information in the form of messages, records, their structures, movements and the manipulation of their components. MEASUR starts with the analysis of meanings, then it examines the communication acts necessary to cause the norms to be executed, and only then begins to consider the more elaborate forms of information, where conventional analysis begins. MEASUR goes behind the normal starting point for requirements engineering to the norms that define the exact, minimal conditions that must be met by every satisfactory IT system, whatever messages, records and data-processing are adopted. That the designer should be free to propose any solution that does not violate the absolutely essential requirements, is a well-established principle. This approach takes a step further back than has been possible before, giving the designer greater scope for business re-engineering before moving towards software engineering. Moreover, the specification of minimal requirements in terms of norms and meanings are succinct and, being expressed entirely in business terms, are easy for users to understand. In addition, the formal representation of these norms explains why business requirements expressed this way should be an appropriate basis for expressing software requirements, too. Organisational semiotics (Andersen 1997, Holmqvist 1996, Liu 1993 and 2000, Liu et al. 2000, Stamper 1973), the theory of signs for organised activities, has important theoretical advantages. As computer programs employ shared data-structures so people need to share perceptual norms relating their knowledge to a certain business activity. Perceptual norms can be understood in terms of invariant patterns of human behaviour. By defining the meanings of words, numbers and other expressions in terms of the behaviours they stand for, we arrive a powerful, new logical concept – that of Ontological Dependency (Stamper, 1985) - leading to a semantic schema with valuable practical properties. Foremost of these is that it imposes on perceptual norms a canonical structure, or Semantic Normal Form (SNF) (Ades 1999, Stamper 1988 & 2000). The SNF consists of atomic, semantic building blocks that contain a minimal set of properties in a standard structure. This standard structure can be used as a default basis for designing prototypes of computer-based information systems. Each semantic atom has the same uniform structure including, among others, two time attributes (start and finish events) linked to responsible human agents. Validating the SNF plays a key role in this research, as does a semantic dictionary for making the semantic atoms and their canonical structures available to the analysts. With this and other tools we expect to be able greatly to reduce the time and specialist skills needed for business analysis. MEASUR comprises three main sets of techniques for:
The hypothesis In the broadest terms: the organisation = the information system Hence the method precisely and formally specifies an organisation’s requirements in terms of meanings, intentions, authorities, responsibilities, and time parameters, as the basis for IT-independent specification. Users can easily verify the specifications as they only contain business concepts; the designs based on the SNF are stable, so easy to maintain. Their formality and rigour make them easy to implement, in part by automatic translation from the business specification. It is important to emphasise that this broad hypothesis is far removed from the intuitive ‘soft systems’ tradition. It invites rigorous testing by the principle of attempted, empirical refutation of a series of particular, more tightly formulated hypotheses. The earlier work outlined above has been carried out in the spirit of Popper’s Refutationist Methodology. Although the theory and methods do have an unusual philosophical foundation, they will be subject to examination by strict scientific and engineering criteria: Do they work correctly, robustly and economically in practice? Aims and Objectives The overall aim of this project is to complete the research programme to develop a new methodology for requirements analysis and specification (MEASUR), based on the concepts of Organisational Semiotics. Many separate techniques have been developed over the last 20 years and validated individually. This research will:
The concrete objectives that define the work packages in the project are:
It is important to add that we are conscious of the needs a) to relate these new methods to established methodologies in which firms have invested so much and b) other, on-going, related research in this problem domain. References Ades, Y. M., 1989, NAMAT System User Manual. SIS project, University of
Quatar. |