Results 11 -
13 of
13
Abstract Selection Games and Deterministic Lotteries
"... The design of deterministic and fair mechanisms for selection among a set of self-motivated agents based solely on these agents ’ input is a major challenge for electronic commerce. These mechanisms are a special case of zerosum games where the only possible outcomes are selections of a single agent ..."
Abstract
- Add to MetaCart
(Show Context)
The design of deterministic and fair mechanisms for selection among a set of self-motivated agents based solely on these agents ’ input is a major challenge for electronic commerce. These mechanisms are a special case of zerosum games where the only possible outcomes are selections of a single agent among the set of agents. We assume the lack of an external coordinator, and therefore we focus on mechanisms which have a solution where the agents play weakly dominant strategies. Our first major result shows that dominated strategies could be added to any selection mechanism, so that the resulting mechanism becomes quasi-symmetric. For fairness, we require the mechanism to be non-imposing; that is, the mechanism should allow any agent to be selected in such a solution. We first show that such mechanisms do not exist when there are two or three agents in the system. However, surprisingly, we show that such mechanisms exist when there are four or more agents. Moreover, in our second major result, we show that there exist selection mechanisms that implement any distribution over the agents, when the agents play mixed dominant strategies. These results also have significance for distributed computing, ranking systems, and social choice. 1
Some Institution
, 2008
"... We carefully investigate humanity’s intuitive understanding of trust and extract from it fundamental properties that succinctly synthesize how trust works. From this detailed characterization we propose a formal, complete and intuitive definition of trust. Using our new definition, we prove simple p ..."
Abstract
- Add to MetaCart
(Show Context)
We carefully investigate humanity’s intuitive understanding of trust and extract from it fundamental properties that succinctly synthesize how trust works. From this detailed characterization we propose a formal, complete and intuitive definition of trust. Using our new definition, we prove simple possibility and impossibility theorems that dispel common misconceptions, expose unexplored areas in the design of reputation systems and shed new light on the shortcomings of previous impossibility results.
1Incentive Compatible Influence Maximization On Social Networks
"... Abstract—Information diffusion and influence maximization on social networks are well studied problems and various models and algorithms have been proposed. The main assumption in these studies is that the influence probabilities are known to the social planner. The influence probabilities can vary ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract—Information diffusion and influence maximization on social networks are well studied problems and various models and algorithms have been proposed. The main assumption in these studies is that the influence probabilities are known to the social planner. The influence probabilities can vary significantly with the type of the information and the time at which the information is recommended. Also there is no way to measure these probabilities from available data on social networks. The most accurate sources to obtain influence probabilities are the influencers themselves. In this report we formulate a game theoretic model of the information diffusion process so as to elicit influence probabilities truthfully from the agents. We present some preliminary results on applying mechanism design to the important problem of influence maximization in the context of information diffusion in social networks.
