Sensible Graph Analytics for Massive Interlinked Data and Social Networks

来自世界各地数百万人、大规模开放数据运动、科学实验、商业信息系统、社交媒体等的数据以指数速度生成。这些数据通常表示为将实体(人、组织和对象)彼此链接在一起的事实网络。此外，由于不断增加的数据集成倡议，不同数据源中的实体相互链接在一起。结果是大量的数据被表示为图(网络)，其中节点是实体，边是它们之间的链接。*图也是在线社交网络建模和信息传播的基础。社交媒体和网络可以接触到全球数十亿用户，并在很短的时间内传播重要信息。大多数企业和其他组织都了解社交媒体和网络的巨大力量；然而，找到最有效的方式来挖掘这一潜力仍然是一个难以实现的目标。*我们研究的统一主题和新颖性是同时应对现实生活中的挑战，即数据链和社会关系的数量、不确定性和不精确性。在这些条件下高效地计算可靠的图分析变得困难得多。*我们努力回答的一些问题是：如何在不确定和不精确的情况下从大型图形库中提取有洞察力的、高置信度的模式？如何在不确定的影响模式和竞争信息提供商的影响下，在大型社交网络中找到有影响力的个人和社区？如何限制“坏”信息的传播，加速“好”信息的传播？如何利用现代系统进行海量计算，以解决在分析大量相互关联的数据和社会网络时的计算挑战？*我们的研究有可能对科学、工程和商业的许多不同领域的大量数据分析活动产生重大影响。**

Data is generated at an exponential rate from millions of people around the world, large-scale open-data movements, scientific experiments, business information systems, social media, and so on. This data is often represented as a web of facts that link entities (people, organizations, and objects) to each other. Also, entities in different data-sources are interlinked together due to ever-increasing data integration initiatives. The result is an enormous amount of data which is represented as graphs (networks), where the nodes are entities and the edges are the links between them. ***Graphs are also the backbone of modeling online social networks and diffusion of information. Social media and networks can reach out to billions of users worldwide and spread out vital information in a very short amount of time. Most businesses and other organizations have understood the tremendous power of social media and networks; however, finding the most effective ways to tap into this potential is still an elusive goal.***The unifying theme and novelty of our research is in addressing simultaneously the real life challenges of volume, uncertainty, and impreciseness of data links and social connections. Efficiently computing reliable graph-analytics under these conditions becomes much harder. ***Some of the questions we strive to answer are: How to extract insightful, high-confidence patterns from large graph databases under uncertainty and impreciseness? How to find influential individuals and communities in large social networks under uncertain influence patterns and competitive information providers? How to limit the spread of "bad" information and accelerate the spread of "good" information? How to make use of modern systems for massive computing to solve computational challenges in analyzing very large graphs of interlinked data and social networks? ***Our research has the potential for great impact in a multitude of data analysis activities in many diverse areas of science, engineering, and business.**