International Journal of Information Technology & Decision Making

This paper examines a decision support system (DSS) for the appraisal of complex decision problems using multi-criteria decision analysis (MCDA). The DSS makes use of a structured hierarchical approach featuring the multiplicative AHP also known as the REMBRANDT technique. The paper addresses the influence of the progression factor used when transforming the decision makers' verbal responses from a semantic to a geometric scale in REMBRANDT. Conventionally, the progression factor 2 is used for calculating scores of alternatives and √2 for calculation of criteria weights. Tests are conducted on the magnitude of these progression factors in order to examine the sensitivity towards the final outcome of an analysis. For illustration a case study concerning the appraisal of a large infrastructure project is presented. The results of the sensitivity calculations are compared with the results of a conventional AHP calculation in order to examine what impact the choice of progression factors as well as the choice of technique have on the decision making. Based on this a modified progression factor for the calculation of scores for the alternatives in REMBRANDT is suggested while the progression factor for the criteria weights is suggested to be kept unchanged. Finally, conclusions and perspectives are set out.

The problem of knowledge-based multiattribute classification with nonorderable classes is considered within the Verbal Decision Analysis (VDA) paradigm. Two VDA-based methods for such problem solving — NORClass and STEPCLASS — are outlined. The STEPCLASS method is initially designed for the problem with nonordered values of attributes. The main idea of NORClass is based on the assumption that a domain expert is able to order the values of any attribute according to their typicality for each class differently and independently of the values of other attributes. We propose to integrate into the STEPCLASS method the main ideas from NORClass. It is shown that such integration allows increasing the efficiency of the STEPCLASS method in cases, where the above assumption is true, and to overcome some drawbacks of NORClass.

In a real-world production problem, some of the parameters involved, namely the availability of the resources, technological coefficients and the incomes associated with each product, may not be perfectly determined thus reflecting some ambiguity or fuzzy understanding of their nature. Therefore these parameters may admit a representation as fuzzy numbers. In this paper, we show how multi-criteria decision theory can be used to deal with a fuzzy linear production problem and with the associated fuzzy linear production game.

Knowledge resource reuse is becoming a widespread approach in the ontology engineering field because it can speed up the ontology development process. In this context, the NeOn Methodology specifies some guidelines for reusing different types of knowledge resources (ontologies, nonontological resources, and ontology design patterns). These guidelines prescribe how to perform the different activities involved in any of the diverse types of reuse processes. One such activity is to select the best knowledge resources for reuse in an ontology development. This selection activity is a complex decision-making problem involving conflicting objectives, like understandability, integration or reliability. We propose a multi-attribute utility theory (MAUT) approach to deal with the selection of the best domain ontologies for reuse, stressing the identification of attributes to measure ontology performances. We take advantage of the sensitivity analysis tools provided by the GMAA system, a PC-based decision support system based on an additive multi-attribute utility model, to exploit imprecise information on the inputs. An example concerning the selection of a subset of ontologies for reuse in the development of a new ontology in the sports domain illustrates the approach.

We develop an interactive partitioning approach for solving the multiobjective decision making problem of a decision maker (DM) who has an implicit general monotone utility function. The approach reduces feasible solution space using the DM's preferences. Hypothetical solutions called partition ideals (PIs) that dominate portions of the efficient frontier are generated and those that are inferior to a feasible solution are used to eliminate the dominated regions. We investigate the issues in representation of the reduced feasible solution space. We develop procedures for locating PIs and measuring the size of feasible solution space. We incorporate these ideas into an approach that converges to a neighborhood of the most preferred solution of the DM. We demonstrate the approach on an example problem.

Water management has become a challenging problem worldwide, especially in developing countries, due to several reasons including the growing scarcity of natural water resources; a demographic explosion in many urban settlements, followed by a rapid urbanization and industrialization of neighboring areas; and the inherent increase in the demand of energy and natural fresh water for human consumption and agricultural developments. The Kwanza river, the longest and most plentiful in Angola, is a paradigmatic case, as it is one of the main sources of wealth in that country. Many people base, direct or indirectly, their living on its water, and many economical activities, including energy production, depend also on its stream. We provide a model for the multiobjective multipurpose management of the Kwanza river. The problem is complicated by the need to take into account uncertainty in various involved processes, to plan over a long period of time, and the presence of several conflicting objectives. We describe also a decision support system implementing our model.

A Markov decision process (MDP) is a general model for solving planning problems under uncertainty. It has been extended to multiobjective MDP to address multicriteria or multiagent problems in which the value of a decision must be evaluated according to several viewpoints, sometimes conflicting. Although most of the studies concentrate on the determination of the set of Pareto-optimal policies, we focus here on a more specialized problem that concerns the direct determination of policies achieving well-balanced tradeoffs. To this end, we introduce a reference point method based on the optimization of a weighted ordered weighted average (WOWA) of individual disachievements. We show that the resulting notion of optimal policy does not satisfy the Bellman principle and depends on the initial state. To overcome these difficulties, we propose a solution method based on a linear programming (LP) reformulation of the problem. Finally, we illustrate the feasibility of the proposed method on two types of planning problems under uncertainty arising in navigation of an autonomous agent and in inventory management.

This paper describes a new multi-stage technique PAKS (the abbreviation of Russian words: Consequent Aggregation of Classified Situations) for sorting multi-attribute objects by a group. Based on decision maker preferences, this technique provides a hierarchical aggregation of a large number of initial attributes into a smaller number of criteria, thus reducing the dimension of the attribute space progressively, using various tools of verbal decision analysis. The consequent aggregation of attributes allows generating manifold lists of composite criteria with numerical and/or verbal scales, finding and analyzing the most preferable solution of a decision problem. The technique was applied to evaluate efficiency of R&D projects, which were assessed by several experts on many qualitative criteria and subsidized by the Russian Foundation for Basic Research.

Investment in stock market involves many decision criteria and variables; hence investors are increasingly relying on the ratings provided by rating agencies to guide their stock selections. However, do these stock ratings have information value? Are these agencies' ratings valid? We establish the dominance cone principle and use the Value Line stock ratings to demonstrate the application of the dominance principle. Our results based upon limited data show that the Value Line rating does not support the notion that better rating results in better rate of return during 2006–2007.