Alexandru Liviu Olteanu

Formalizing and solving the problem of clustering in MCDA

The topic of clustering has been widely studied in the field of Data Analysis, where it is defined as an unsupervised process of grouping objects together based on notions of similarity. Clustering in the field of Multi-Criteria Decision Aid (MCDA) has seen a few adaptations of methods from Data Analysis, most of them however using concepts native to that field, such as the notions of similarity and distance measures. As in MCDA we model the preferences of a decision maker over a set of decision alternatives, we can find more diverse ways of comparing them than in Data Analysis. As a result, these alternatives may also be arranged into different potential structures. In this paper we wish to formally define the problem of clustering in MCDA using notions that are native to this field alone, and highlight the different structures which we may try to uncover through this process. Following this we propose a method for finding these structures. As in any clustering problem, finding the optimal result in an exact manner is impractical, and so we propose a stochastic heuristic approach, which we validate through tests on a large set of artificially generated benchmarks.

A clustering approach using weighted similarity majority margins

We propose a meta-heuristic algorithm for clustering objects that are described on multiple incommensurable attributes defined on different scale types. We make use of a bipolar-valued dual similarity-dissimilarity relation and perform the clustering process by first finding a set of cluster cores and then building a final partition by adding the objects left out to a core in a way which best fits the initial bipolar-valued similarity relation.

Towards a Protocol for Inferring Preferences Using Majority-rule Sorting Models

In Multi-Criteria Decision Aiding, one of the current challenges involves the proper integration and tuning of the preference models in real-life contexts. In this article, we consider the multi-criteria sorting problem where the decision makers preferences fall within the outranking paradigm. Following recent advances on extensions of classical majority-rule sorting models, we propose a methodology for adapting them to the perspective of the decision maker. We illustrate the application of the methodology on a real-world problem linked to the evaluation of contributors within Free/Libre Open Source Software communities. The experiments that we have carried out show that the various considered model extensions appear to be useful from the perspective of decision makers in a real-life preference elicitation process, and that the proposed methodology gives useful indications that can serve as guidelines for analysts involved in other elicitation processes.

MODEL: a multicriteria ordinal evaluation tool for GIS

While the integration of geographical information systems (GIS) and multicriteria decision aiding (MCDA) has attracted increasing interest from researchers in recent years, due to the wide array of applications that can benefit from GIS as well as the different types of decision problems and various models that can be used through MCDA, plenty of opportunities of integrating GIS and MCDA remain. In this article, we present the result of such an opportunity in the form of a methodology and software that is currently used by the Naval Hydrographic and Oceanographic Service in France. Furthermore, this tool may be used in conjunction with other GIS–MCDA applications with a single decision maker, multiple decision makers or even where the decision has a hierarchical structure.

Descriptive Profiles for Sets of Alternatives in Multiple Criteria Decision Aid

In the context of Multiple Criteria Decision Aid, a decision-maker may be faced at any time with the task of analysing one or several sets of alternatives, irrespective of the decision he is about to make. As in this case the alternatives may express contrasting gains and losses on the criteria on which they are evaluated, and while the sets that are presented to the decision-maker may potentially be large, the task of analysing them becomes a difficult one. Therefore the need to reduce these sets to a more concise representation is very important. Classically, profiles that describe sets of alternatives may be found in the context of the sorting problem, however they are either given beforehand by the decision-maker or determined from a set of assignment examples. We would therefore like to extend such profiles, as well as propose new ones, in order to characterise any set of alternatives. For each of them, we present several approaches for extracting them, which we then compare with respect to their performance.