"The Application and Analysis of Social Structures in Trust, Influence and Search Problems in Agent Societies"

了解信息在社会和经济网络中传播的影响和后果在过去几年里引起了越来越多的关注。最初的研究集中在结构网络的性质上，然而，现在正在研究的问题中考虑了代理行为的重要性和信息在网络中流动的语义。对这类网络的研究在生物信息学、系统生物学、社会科学和电子商务等广泛领域都具有相当重要的意义。 这项研究关注的是动态的社会和经济网络，即网络中的拓扑和信息流随着时间的变化。具体地说，我们感兴趣的是构建基于代理的模型，在该模型中，代理与其直接邻居交互，并根据它们的交互选择添加或删除邻居。当考虑这样的模型时，自然会出现几个问题。首先，我们如何选择要与之交互的新代理？其次，在什么情况下，我们才能停止与代理人的互动？第三，网络中的特定代理的影响力有多大，需要哪组代理才能确保网络中的意见或代理状态收敛到特定值？最后，如何利用代理人的社交网络回答问题或提出建议？这些问题只是例子，代表了我们整体研究愿景的一部分。我们研究的一个目标是在分布式或基于社交的搜索算法领域做出贡献。例如，通过了解社交网络中影响力的本质，我们有望设计出在电子商务中有重要应用的改进的推荐系统。

Understanding the effects and consequences of information diffusion in social and economic networks has attracted increasing attention over the last few years. Initial investigations focused on structural network properties; however, the importance of agent behavior and the semantics of information flowing in the network is now being considered in the research problems investigated. The study of such networks is of considerable importance in a wide range of fields; e.g., bioinformatics, systems biology, social science and electronic commerce. The research proposed concerns itself with dynamic social and economic networks; i.e., networks in which the topology and information flows change with time. Specifically, we are interested in constructing agent-based models in which agents interact with their immediate neighbors and choose to add or remove neighbors based upon their interactions. Several questions naturally arise when considering such models. First, how do we select new agents with which to interact? Second, under what conditions do we cease to interact with an agent? Third, how influential is a particular agent in the network and what set of agents would be required to ensure opinion, or agent state, in the network converges to a specific value? Finally, how can an agent's social network be used to answer questions or make recommendations? These questions are examples only and represent a part of our overall research vision. A goal of our research would be to make contributions in the area of distributed, or socially-based, search algorithms. By understanding the nature of influence in social networks we would expect to be able to design improved recommender systems which have important applications in e-commerce, for example.