Using soft computing to build real world intelligent decision support systems in uncertain domains

https://doi.org/10.1016/S0167-9236(00)00135-4Get rights and content

Abstract

Whilst the builders of traditional decision support systems have regularly used game theory and operations research, they have rarely used statistical techniques to build intelligent support systems for fields that have weak domain models. Further, the principle tools in the artificial intelligence arsenal were centred on symbol manipulation and predicate logic, while the use of numerical techniques were looked upon with disfavour.

We claim that soft computing techniques (such as fuzzy reasoning and neural networks) can be integrated with symbolic techniques to provide for efficient decision making in knowledge-based systems. We illustrate our claim through the discussion of two decision support systems that have been constructed using soft computing techniques. Split-Up uses rules and neural networks to advise on property distribution following divorce in Australia, whilst IFDSSEA uses fuzzy reasoning to assists teachers in New York State to grade essays.

We focus on how both systems reason and how they have been evaluated.

Section snippets

An introduction to decision support systems and soft computing

According to Turban and Aronson [32] a decision support system (DSS) is a computer-based information system that combines models and data in an attempt to solve non-structured problems with extensive user involvement. They claim an expert system (ES) is a computer system that applies reasoning methodologies on knowledge to render advice or recommendations much like a human expert. When expert systems technology was first applied to decision-making problems, it fell short in several respects.

Reasoning in discretionary domains

Black [6] claims that discretion is a power or right conferred upon decision-makers to act according to the dictates of their own judgment and conscience, uncontrolled by the judgment or conscience of others. Nevertheless, decision-makers must in accordance with the rule of law and their decisions must preserve the rights of all parties effected by the decision-making. It is thus essential that discretionary decision making not be arbitrary—since an arbitrary application of discretion could

The Split-Up system: reasoning and learning in domains characterised by uncertainty

KDD is an emerging field combining techniques from databases, statistics and artificial intelligence, which is concerned with the theoretical and practical issues of extracting high level information (or knowledge) from a large volume of low-level data. Fayyad et al. [11] define knowledge discovery in databases (KDD) as ‘the non-trivial process of identifying valid, novel, potentially useful understandable patterns in data’. Because most KDD systems use some form of statistical algorithm to

Current research involved in developing Split-Up into a commercial tool

In building negotiation support tools we have assumed that all actors behave rationally. Principled negotiation [13] promotes deciding issues on their merits rather than through a haggling process focused on what each side says it will and will not do. It promotes a focus on interests of the party rather than allowing negotiation to deteriorate into a contest of ‘who will back down first’. Fundamental to the concept of principled negotiation is the notion of Know your best alternative to a

Intelligent fuzzy decision support systems that provide advice about assessment

Along with evaluation and interpretation, assessment is a fundamental decision making task. Examples of assessment tasks involve evaluating student performance and the viability of various investment options. Often decisions need to be made with insufficient numerical data, or imprecise or vague information [16].

Grading essays is labour intensive, repetitive, and fraught with imprecision. Typically a teacher must learn a scoring standard or ‘rubric’ that he or she will consistently apply to all

Evaluating intelligent soft computing decision support systems

An important feature in assessing the value of intelligent decision support systems is to assess their outcomes. Until recently, very little emphasis has been placed on evaluating intelligent systems. Recently, artificial intelligence has become an empirical science. Most of the research being conducted under the SPIRT grant discussed in Section 4 is dedicated to evaluating legal expert systems. [23], [24] and the Seventh International Conference on Artificial Intelligence and Law [23] focus on

Conclusion

Until recently, the principle tools in the artificial intelligence arsenal were centred on symbol manipulation and predicate logic, while the use of numerical techniques were looked upon with disfavour. What is more obvious today is that symbol manipulation and predicate logic have serious limitations in dealing with real world problems in the realm of decision making. In this paper, we have focused on how soft computing techniques—in particular fuzzy logic and neural networks—can help build

John Zeleznikow is the Director of the Donald Berman Laboratory for Information Technnology and Law at the Applied Computing Research Institute, La Trobe University. His research interests are in the areas of intelligent decision support systems, knowledge discovery from databases, and hybrid reasoning. He has published two books on artificial intelligence and law.

References (45)

  • J.M. Bonham

    Cognitive mapping as a technique for supporting international negotiation

    Theory Decision

    (1993)
  • Buchanan et al.

    An intelligent interactive system for delivering individualised information to patients

    Artif. Intell. Med.

    (1995)
  • J. Diederich

    Explanation and artificial neural networks

    Int. J. Man-Mach. Stud.

    (1992)
  • R.C. Eberhart

    Using evolutionary computation tools in explanation facilities

    Int. J. Expert Syst.

    (1995)
  • Fayyad et al.

    The KDD process for extracting useful knowledge from volumes of data

    Commun. ACM

    (1996)
  • Fisher et al.

    Getting to YES: Negotiating Agreement Without Giving In

    (1981)
  • Gorry et al.

    A framework for management information systems

    Sloan Manage. Rev.

    (1971)
  • B. Kosko

    Fuzzy Thinking: The New Science of Fuzzy Logic

    (1993)
  • Nolan

    A conceptual model for an intelligent fuzzy decision support system

    Heuristics

    (1997)
  • Nolan

    An intelligent system for case review and risk assessment in social services

    AI Mag.

    (1998)
  • O'Leary et al.

    Validating expert systems

    IEEE Expert

    (1990)
  • Skabar et al.

    Using argumentation for the decomposition and classification of tasks for hybrid system development

  • Cited by (0)

    John Zeleznikow is the Director of the Donald Berman Laboratory for Information Technnology and Law at the Applied Computing Research Institute, La Trobe University. His research interests are in the areas of intelligent decision support systems, knowledge discovery from databases, and hybrid reasoning. He has published two books on artificial intelligence and law.

    James Nolan is a professor in the Quantitative Business Analysis and Computer Science departments at Siena College. He holds a PhD from the Rockefeller College at the University of Albany. His research interests are in the areas of statistical and machine learning, hybrid intelligent systems design and development, fuzzy decision support systems, and neural networks. He has published numerous articles in these areas.

    View full text