Paper Type
Complete Research Paper
Description
Decision making in operational planning is increasingly affected by conflicting interests of different stakeholders such as subcontractors, customers, or strategic partners. Addressing this, automated negotiation is a well-suited mechanism to mediate between stakeholders and search for jointly beneficial agreements. However, the outcome of a negotiation is strongly dependent on the applied negotiation protocol defining the rules of encounter. Although protocol design is well discussed in literature, the qustion on which protocol should be selected for a given scenario is little regarded so far. In this study, we propose a decision support system for negotiation protocol selection (DSS-NPS) that is based on a machine learning approach "“ an artificial neural network (ANN). For evaluation purposes, we trained the ANN by simulating millions of intercompany machine scheduling negotiations. By using observable and revealed characteristics, the ANN can achieve a 58% smaller prediction error compared to a linear regression. The proposed protocols of the DSS-NPS realize negotiation outcomes (measured as the average level of satisfaction) that are significantly better and more robust than results based on a regression or the best protocol of the simulations (p-Valus: 0.049% and 0.026%). Concluding, the proposed DSS-NPS represents a beneficial artefact for finding adequate protocols dynamically.
DECISION SUPPORT FOR NEGOTIATION PROTOCOL SELECTION: A MACHINE LEARNING APPROACH BASED ON ARTIFICIAL NEURAL NETWORKS
Decision making in operational planning is increasingly affected by conflicting interests of different stakeholders such as subcontractors, customers, or strategic partners. Addressing this, automated negotiation is a well-suited mechanism to mediate between stakeholders and search for jointly beneficial agreements. However, the outcome of a negotiation is strongly dependent on the applied negotiation protocol defining the rules of encounter. Although protocol design is well discussed in literature, the qustion on which protocol should be selected for a given scenario is little regarded so far. In this study, we propose a decision support system for negotiation protocol selection (DSS-NPS) that is based on a machine learning approach "“ an artificial neural network (ANN). For evaluation purposes, we trained the ANN by simulating millions of intercompany machine scheduling negotiations. By using observable and revealed characteristics, the ANN can achieve a 58% smaller prediction error compared to a linear regression. The proposed protocols of the DSS-NPS realize negotiation outcomes (measured as the average level of satisfaction) that are significantly better and more robust than results based on a regression or the best protocol of the simulations (p-Valus: 0.049% and 0.026%). Concluding, the proposed DSS-NPS represents a beneficial artefact for finding adequate protocols dynamically.