Group Decision and Negotiation最新文献

Learning an opponent’s preferences in bilateral multi-issue automated negotiations can lead to more favorable outcomes. However, existing opponent models can fail in negotiation contexts when their assumptions about opponent behaviors differ from actual behavior patterns. Although integrating broader behavioral assumptions into these models could be beneficial, it poses a challenge because the models are designed with specific assumptions. Therefore, this study proposes an adaptable opponent model that integrates a general behavioral assumption. Specifically, the proposed model uses Bayesian learning (BL), which can apply various behavioral assumptions by considering the opponent’s entire bidding sequence. However, this BL model is computationally infeasible for multi-issue negotiations. Hence, current BL models often impose constraints on their hypothesis space, but these constraints about the utility function’s shape significantly sacrifice accuracy. This study presents a novel scalable BL model that relaxes these constraints to improve accuracy while maintaining linear time complexity by separately learning each parameter of a utility function. Furthermore, we introduce a general assumption that the opponent’s bidding strategy follows a concession-based pattern to enhance adaptability to various negotiation contexts. We explore three likelihood function options to implement this assumption effectively. By incorporating these options into the proposed scalable model, we develop three scalable concession-driven opponent models using Bayesian learning (COMB). Experiments across 45 negotiation domains using 15 basic agents and 15 finalists from the automated negotiating agents competition demonstrate the proposed scalable model’s higher accuracy than existing scalable models. COMB models show higher adaptability to various negotiation contexts than state-of-the-art models.

This work proposes and studies a dynamic model of two bargaining parties exchanging offers over time, considering their confidence about the share of the “pie” they obtain, which translates into expectations regarding the outcome of the bargaining process. The model predicts the sequence of offers as well as the final agreement for given confidence parameters. A mathematical analysis of the model shows the outcome is an Asymmetric Nash Bargaining Solution with exponents determined by the bargainers’ confidence. Moreover, a compensation effect can be found between confidence and risk aversion. This work also considers that confidence levels of bargainers might change during the negotiation, and we conduct a comprehensive simulation study to analyze the effect of such changes. Through Monte-Carlo simulation, we show that a bargainer is better off if its confidence increases, but the advantage is lost if the other party’s confidence increases in a similar way. In that case, concessions are smaller and negotiations last longer. Changing confidence parameters make the outcome harder to predict, as it will depend more on the final confidence than the initial one. The simulations also show that the average size of concessions, and therefore the final agreement, depend not only on whether confidence increases or decreases, but also on the change rate, with stronger effects observed when change accelerates towards the end of the process.

In the literature, many desirable properties for allocations of indivisible goods have been proposed, including envy-freeness, Pareto optimality, and maximization of either the total welfare of all agents, the welfare of the worst-off agent, or the Nash product of agents’ welfares. In the two-person context, we study relationships among these properties using both analytical models and simulation in a setting where individual preferences are given by additive cardinal utilities. We provide several new theorems linking these criteria and use simulation to study how their values are related to problem characteristics, assuming that utilities are assigned randomly. We draw some conclusions concerning the relation of problem characteristics to the availabilty of allocations with particular properties.

Consensus-reaching among decision-makers (DMs) is an important prerequisite for effective group decision-making. Determining a collective value function that is recognized by major DMs is new in consensus research. We are approaching this problem by adopting the preference disaggregation analysis (PDA) to construct a novel consensus-reaching process (CRP). More precisely, we define the value function that can restore the preference information of all DMs as the consensus value function, and determine all such value functions by the PDA method. A consensus discriminant model is constructed to determine whether DMs can reach a consensus. Considering the adjustment cost of DMs, the minimum cost consensus model, and the minimum cost inconsistency elimination model, are constructed by introducing estimation errors and 0–1 variables, respectively, thus assisting DMs to reach a consensus. Furthermore, in the process of selecting a representative collective value function from the consensus space for subsequent decision analysis, a lexicographic optimization process is applied to convert the multi-objective programming problem of DMs’ individual requirements for the collective value function into a multi-stage single-objective programming problem. This study provides a new concept of consensus and extends the classic minimum cost consensus model to the case of value functions. Finally, an illustrative example showing the proposed CRP in action is presented, while conducting sensitivity analysis to explore the impact of parameter changes on the model.

In a decision-making problem where a group must select an action from risky lotteries to receive a payoff that needs to be distributed among the group members, the group’s external risky choice and internal resource allocation emerge as two critical and interrelated subproblems. Individuals unconsciously exhibit social comparison behavior in many group contexts, profoundly impacting their payoff preferences. In this study, we first formulate a novel group resource allocation model and explore how the allocation equality of group resource is influenced by social comparisons. Particularly, we discuss the non-dictatorship condition under social comparison effects, which guarantees an extreme case of group resource allocation, i.e., “winner-take-all,” does not appear. Subsequently, we investigate the group risky choice under the effects of social comparison. The main results show that (1) introducing social comparison effects can increase allocation equality when loss aversion is high but decrease allocation equality when loss aversion is low, and (2) the classical risk sharing rule still holds in the group under social comparison effects, but introducing social comparison effects will lead to the group being more risk averse.

In a ranked voting system, voters select a subset of candidates and rank them from most to least preferred. Data envelopment analysis (DEA)-based voting models, among others, are used to determine the rank-position weights most favorable for each candidate, with the goal of achieving the highest aggregate score. However, concerns have been raised about the weights assigned to each rank position, as well as the potential for rank reversal of some candidates resulting from changes in votes earned by other candidates. To address these issues, some authors have developed two improved models. These models aim to incorporate the constraints of candidates that are not being evaluated into a single restriction, preventing inefficient candidates from influencing the order of efficient candidates. Moreover, these models treat the parameters used to make the distance between successive ranks as variable weights, and calculate average efficiency scores of candidates while considering the entire range of parameters. In this study, we revisit the two improved models and explore an alternative approach based on results from linear algebra and convex analysis, which is more intuitive and easier to understand. Furthermore, we provide closed-form optimal solutions for DEA-based voting models that share the common goal of maximizing the distance between successive ranks while considering both efficiency-related and weight constraints. The analysis of these four models offers a better understanding of their similarities and differences.

This study presents empirically calibrated simulations of three different variants of environmental matching agreements aimed at reducing global greenhouse gas emissions. We determine whether matching agreements can produce larger stable coalitions and increase abatement contributions and payoffs as compared to standard agreements. The matching agreements we analyze feature uniform matching rates by which coalition members match the unconditional contributions of (i) the other coalition members, (ii) all other players, or (iii) only non-members, while non-members do not commit to any matching and maximize their individual payoffs. The simulation considers twelve asymmetric world regions with linear abatement benefits and quadratic costs, calibrated based on the STACO 3 model, and uses emissions data from the shared socioeconomic pathways database. We find that the first variant of the matching game fails to produce any stable coalitions and thus performs worse than the standard agreement that produces a stable two-player coalition. The second variant produces a stable grand coalition and significantly increases the abatement and payoff levels beyond the non-cooperative Nash baseline. Partial coalitions are unstable in this game. The third variant produces a two-player coalition similar to the standard coalition formation game, but with different members and higher abatement and payoff levels due to the matching mechanism.