An Integrated Approach to Multi-Agent Multi-Criteria Conflict Resolution with Applications to Environmental Management

Liu Ying

Abstract

In recent years, due to the increasing needs for analysing complex real-world problems, practical efforts have been putting into finding possible ways of the integration of analytic approaches in multi-criteria decision-making (MCDM) and multi-agent conflict resolution (MACR) fields. Multiple-criteria decision-making (MCDM) helps individual decision makers (DMs) find optimal alternatives, while multi-agent conflict resolution (MACR) helps DMs understand their opponents and co-players. Numerous works have been done in the two individual fields, but the overlaps of the two fields have not been addressed with widely recognized and well appreciated approaches. This paper is concerned with addressing complex multi-criteria decision-making problems involving multiple agents with conflicting interests.

Common characteristics of issues addressed in this paper include interdependence among decision factors, conflicts of interests among different stakeholders and unequal decision power, which makes such problems troublesome for decision-makers to analyse. Current methodologies are either oversimplified, solving only part of the problem, or tedious and disjoint, which creates problems for manipulation when applied to more complex problems.

One of the main contribution of the methodology proposed in this paper is that it suggests a way to bridge the MCDM and conflict resolution processes in a logical and efficient manner that takes into consideration the multi-lateral influence power among the DMs. Especially for the conflict resolution part, instead of assessing payoff state-by-state and do the vague comparisons, the methodology suggested in this paper enables the decision-makers to study the conflict resolution with clear and direct comparisons and inputs. This ensures more logical and justifiable results.

The methodology proposed in this paper has six steps and can be divided into three functional stages. The first stage is the individual MCDM analysis, when ANP is used to help each DM identify its preference for all available alternatives. The second stage is the influence power analysis, when a newly proposed algorithm is used to join the MCDM results and the influence power comparison results to calculate the state-payoff that can be used directly as inputs for the conflict resolution in the third stage.

Two case studies are analysed using the proposed methodology in this paper. The first case study considers the debated on the Singapore Cross Island Line project, while the second case study tackles the 2015 Southeast Asian haze issues. In the first case, there is a major DM who possesses comparatively large influence power and the right to make the final decision. Final resolution is given by the equilibrium state with the highest state payoff from the major DM’s standpoint. In the second case study, the conflicts between countries lead to more complex situations. As there is no single DM possessing sufficiently large decision power, an inverse GMCR analysis has also been done to help identified the impacts of potential alternations of actions on the final resolution result.

For both of the two cases, the obtained results are compared with real-world resolutions for discussion and for recommendations for future improvements to be given. In addition, comparisons between the two case studies are done to provide more useful insights into the applications of the proposed methodology on real-world environmental management issues. The proposed methodology well solves the multi-agent conflict resolution problems while being able to thoroughly consider the deciding factors and the interdependences in the MCDM analysis. It has also effectively reduced the number of inputs required when there are more stakeholders and alternatives.