Business Systems and Knowledge Modelling Lab

Research Project

Ontology of Information Systems

Research Goal
The foundations of the task based human-centred ontology are underpinned in enabling theories in philosophy (semiotic theory), cognitive science (distributed and situated cognition), workplace (work-oriented design), and psychology (activity theory) and several human-centred criteria outlined in the 1997 NSF workshop on human-centred systems. The enabling theories facilitate integration of the external (i.e. social and organizational realities) and internal (i.e., subjective realities involving stakeholder goals and incentives) planes of information system development. The integration results in the definition of a human-centred ontology at the conceptual and computational levels of software development.

This project deals with development and implementation of a task based human-centred ontology for information systems.

In particular, by software systems we mean software artefacts in the three areas including databases, information systems (also known as traditional software applications) and knowledge/intelligent systems. The approach to development of the framework is based on human-centredness criteria, which makes the users goals and tasks prime drivers in the system development process. The framework is developed at two levels: conceptual and computational.

At the conceptual level, various pragmatic issues related to human-centred software development are considered. These issues centre around areas related to software systems like software engineering, enterprise modelling, intelligent systems, electronic commerce and others. These pragmatic issues along with enabling theories in work-oriented design, artificial intelligence, psychology and multimedia are used to develop four components of the conceptual framework. At the computational level, various software artefacts (e.g. objects, agents, fuzzy systems) are used in a complementary manner based their intutitive modelling strengths to define multiagent architecture. Likewise, intelligent technologies like expert systems, neural networks, fuzzy systems, and generic algorithms in association with enabling technologies like object and agent employed to realise the characteristics of the framework at the computational level are also addressed. The technologies used not only to take account the need to facilitate the communication of intelligent systems with databases and information systems but also to provide an ability to model each of these three software systems using compatible representations at each stage of their development cycle.

In order to demonstrate applicability of the framework, two real world problems in sales and marketing area namely, Order Processing System, and Electronic Brokerage System have been selected.

Problem Space
The period 1970-2000 can be considered as the information technology era in the evolutionary life cycle of business enterprises. In the 1970’s computer-based systems were used for collecting and managing volumes of data. The main objective was to help speed up data processing with aim to improve work efficiency of business functions of an organisation. The main contributing software systems were databases. A database software provided systems managers with the capability to catalogue, maintain, and control the data so that it made it possible to provide a basic support for general management.

Since the 1980’s computerised information systems have been situated virtually in any organisations. The widespread adoption of information technology in organisations has resulted in a number of application software systems developed to serve various needs of users. These computerised information systems were used mainly for generating meaningful information to support management activities along the business lines as well as the organisational levels of a firm. Thus, in the other words, in the 1980’s information has been recognised as a strategic resource that can be exploited as a means for gaining competitive advantage.

The late 1990’s and beyond can be heralded as the era of globalisation of economies, where knowledge is needed for competing in the world market. The globalisation has gained further momentum with the emergence of internet and e-commerce. Many organisations today are following this trend by shifting from purely information based to knowledge based organisations. Knowledge has now become an important strategic resource to help human deal with the complexity and sheer quantum of information. This strategic resource will not only help people to enhance their own performance (work faster, smarter and more efficiently) but also that of the machine with which they work. The areas like knowledge discovery and data mining (Fayyad, Piatetsky-Shaprio et al. 1996; Edelstein 1998) and soft computing (Nwana and Ndumu 1997; Nwana and Wooldridge 1997) are the latest manifestations of intelligent systems and knowledge systems in general.

The process of system development in each of the three areas of enterprise systems (i.e. databases, information systems, knowledge/intelligent systems, and recently e-commerce systems) has employed a wide variety of modelling methods and technologies. For example, database systems are modelled using data abstraction like entity-relationship diagrams, information systems rely on functional abstraction like data flow diagrams, state transition diagrams, and knowledge/intelligent systems rely on technologies like expert systems, fuzzy-logic, neural networks, and genetic algorithms which model different aspects of human-cognition, brain and evolution. This generally wealthy proliferation of modelling methods and technologies, however, has led to a new problem. In recent years, there is a continuous growing interest within the fields of databases, information systems, and knowledge/intelligent systems toward integrated systems which aims to combine the functionality and technical properties of a knowledge/intelligent system with that of a database and/or information system (Kerschberg 1991; Aamodt and Nygard 1995). In line with this research direction is the pragmatic experience currently learnt from the development of knowledge/intelligent systems which indicates that these systems are frequently required to access to and/or integrate with a database system or an information system (Dillon and Tan 1993). Unfortunately, most of the methods used to model the three system areas have their respective life cycle models, automated support tools and techniques, were developed in isolation from, and were designed to be used independently of, other, methods (Kerschberg 1991; Dillon and Tan 1993; Maes 1994; Khosla and Kitjongthawonkul 1998). Hence, there is the lack of ability to model each of these three software systems using compatible representations at each stage of their development cycle. This makes analysis and design of the integration of these systems difficult. In many cases, the database or software application might already be in place. In other cases, they may be being built within a similar time frame. It is useful in all cases for the conceptual model for each of these different kinds of systems to be based on the same paradigm or a set of consistent methodologies (Dillon and Tan 1993; Khosla and Dillon 1997; Khosla and Kitjongthawonkul 1998). The use of a set of consistent methodologies in each of these systems will make future forward and backward integration with one of the other types of systems more manageable (Khosla and Kitjongthawonkul 1998).

Another concern with the proliferation of the modelling methods is that most of them are technology inspired or technology centred. The prime motivation in modelling a system is then on determining technology models (e.g. object-oriented, neural networks, etc.) to provide a software solution to a human problem. Although the benefits of technology can by no means be underestimated, its underpinning principles of rationality and objectivity are not adequate tools for dealing with social and organisational reality in which the system has to operate. The past decade has seen an increasing emphasis on modelling of complex software systems which are based on synergy between human and the machine. In the 70’s and 80’s information technology has been primarily used by organisations for automation without looking into its psychological and social side effects and the revolutionary impact it has had on the overall nature of workplace activity. In the 90’s the disruptive efforts of information technology have become all too visible. Empirical studies on the impact of systems based on technology-centred approaches on practitioner cognition and performance has ranged from bewilderment to loss of human lives (Perrow 1984; Norman 1988; Norman 1993; Sarter, Woods et al. 1997). These studies have forced the organisations to adopt a more balanced view where information technology and computers have to coexist, rather than necessarily replace people and their activities. Computers and information technology are being deployed based on the incentives they offer to workers in terms of their personal goals as well as the organisational goals. The computer and information technology are now seen as tools that assist people in their day-to-day activities and in breakdown situations rather than as prime drivers which redefine workplace activities and tasks in an organisation. Thus, there is a need for a human-centred rather than a technology-centred approach to information systems development.

As mentioned in the preceding paragraph, a wide spectrum of application systems found in an enterprise can be classified into three areas: databases, information systems, and knowledge/intelligent systems. The problem outlined in previous section has reflected the need for integration and interoperability of these systems. Unfortunately, however, the technology mismatch is the one of the major obstacles today for their integration and interoperability. In addition, research fields like knowledge discovery and data mining have demonstrated that one type of system (i.e. a standard database system with DBMS capabilities) can evolve into another type of system (i.e. an intelligent system) to provide sophisticated intelligent decision support. Thus, there is a need to build information systems which are problem driven and which have capabilities to evolve with time. Such information systems will need to be based on architectures that facilitate use of range of technologies for different tasks and needs which evolve with time. In this scenario, technologies are more likely to be used based on their intuitive modeling strengths

Team Members
Rajiv Khosla, Somkiat Kitjongthawonkul

Publications
[1] R. Khosla, I. Sethi and E. Damiani, Intelligent Multimedia Multi-Agent Systems
– A Human-Centred Approach (includes aspects related to socio-technical systems), Kluwer Academic Publishers, Massachusetts, USA,. October 2000, 333 pages

[2] R. Khosla and S. Kitjongthawonkul, 'A Human-Centred Agent-Based Architecture
for Electronic Brokerag in e-Commerce', in International Journal of Soft Computing, Springer-Verlag, vol 5 no. 3, 2001

[3] S. Kitjongthawonkul and R. Khosla, ‘An Information System Ontology of Task-
Based Problem Solving Adapters, Objects and Agents with Application to Sales and Marketing’, in 15th International Conference on Computer and their Applications,
New Orleans, Louisiana, USA, pp. 108-111, 2000

[4] R. Khosla and S. Kitjongthawonkul, ‘A Human-Centered Integrated Framework
of Agents and Objects for Modeling Data, Information and Knowledge in an Enterprise,’ in Ninth Australian Conference on Information Systems (ACIS’98), Sydney, pp. 311-322, September 1998.