Collaborative Research: Fundamental Processes of Network Exchange

拟议的研究是卡罗莱纳大学（约翰·斯克沃雷茨和大卫·威勒）和爱荷华大学（迈克尔·洛瓦格利亚和巴里·马可夫斯基）研究人员的合作成果。这项工作扩展了一个基于理论的权力和社会交换过程调查项目。网络交换理论（NET）是试图解释交换关系网络中权力发展的几种理论之一。虽然之前的研究已经证明。NET的预测比它的竞争对手更准确，但最近在该理论的一个基本部分——图理论功率指数（GPI）中出现了一个问题，该部分将网络划分为强弱功率。为了解决这一问题，研究者们开发了两种独立的交换网络权力关系预测方法。这为大幅度提高交换研究的速度和质量奠定了基础。通过开发集成的计算机程序，研究人员提出了自动化交换网络的系统构建、比较和分类。两种方法不一致的网络可以从数千个潜在测试网络的进一步分析中剔除。这将使pi能够对这些网络中的资源分布做出准确的预测，并将其与计算机模拟进行比较。发现有问题的测试网络并为实验测试做准备的整个过程可以在几个小时内完成，而以前，合适的测试网络在很大程度上是偶然发现的，这个过程可能需要数年时间。然后，pi建议改进实验方法和统计技术，以研究在解决分类问题中至关重要的一些网络。这些进步将使我们更接近模拟社会中常见的复杂网络中的交换过程。* * * ? ?

9515364 Lovaglia The proposed research is a cooperative effort between researchers at the University of Carolina (John Skvoretz and David Willer) and the University of Iowa (Michael Lovaglia and Barry Markovsky). This work extends a program of theory-based investigations on power and social exchange processes. Network Exchange Theory (NET) is one of several theories that attempt to explain the development of power in networks of exchange relations. While previous research has demonstrated that NET predictions are more accurate than its competitors, a problem recently surfaced in a fundamental part of the theory that classifies networks as strong or weak power, the Graph-theoretical Power Index (GPI). In solving this problem, two independent methods of predicting power relations in exchange networks were developed by the investigators. This set the stage for a dramatic improvement in the sped and quality of exchange research. By developing integrated computer programs, the investigators proposed to automate the systematic construction, comparison and classification of exchange networks. Networks for which the two methods disagree can be culled from further analysis from thousands of potential test networks. This will then enable the PIs to make exact predictions for resource distribution in these networks and compare them to computer simulations. The entire process of discovering problematic test networks and preparing them for experimental test can be accomplished in a matter of hours, whereas previously, suitable test networks were discovered largely by chance and the process could take years. Then the PIs propose to improve experimental methods and statistical techniques to investigate a number of networks that were crucial in the solution to the classification problem. These advances will bring us a step closer to modeling exchange processes in complex networks commonly found in society. *** ??