III: Small: Collaborative Research: Mining Information Propagation on the Web

实时信息是Web内容创建和管理中出现的一个基本问题。人们越来越多地参考新闻站点、博客、社交网络站点和Twitter feed上动态变化和迅速传播的信息，而不是参考周期性索引的相对静态来源。这个项目将研究信息内容是如何随时间变化的，它是如何通过潜在的社会网络传播的，以及它的接收者是如何将它组装成更大的单元的。该项目将探索解决这些问题的新技术，基于跟踪、分析和呈现随时间迅速发展和传播的信息的新方法。由此产生的方法旨在改变处理web上实时信息的方式的重要方面。首先，通过底层信息网络传播的基本信息单位将被识别。从近10亿篇新闻媒体文章和博客文章中(6TB的数据)，以及来自Twitter的5亿条tweet的集合，小的（通常是文本的）信息单元将被识别出来，这些信息单元在通过网络传播时保持相对稳定。这些基本单位的时间变化将被分析和建模。这种模型将包括与流行病的生物学模型的联系，以及利用生物和社会传染之间根本差异的新框架。最后，时间变化将与该信息扩散的网络级模型有关。通常，实时信息传播的实际网络无法直接观察到，也无法直接测量网络中任何特定节点的影响。因此，该项目将开发机器学习技术来推断这些隐藏的网络和未被观察到的影响水平。欲了解更多信息，请参阅该项目的网站：http://snap.stanford.edu/proj/mipro

Real-time information is a fundamental emerging issue in the creationand management of Web content. Increasingly, rather than consultingrelatively static sources that are indexed on a periodic basis,people refer to information on news sites, blogs, social-networkingsites, and Twitter feeds that change dynamically and spread rapidly.This project will study how information content varies over time, howit is transmitted through underlying social networks, and how itsrecipients assemble it into larger units. The project will explorenew techniques for addressing these issues, based on novel methodsfor tracking, analyzing, and presenting information that evolves andspreads rapidly over time. The resulting approach aims to transformimportant aspects of the ways in which real-time information on theWeb is handled.First the fundamental units of information that spread through theunderlying information networks will be identified. From a set ofnearly 1 billion news media articles and blog posts (approx. 6TB ofdata), and a collection of 500 million tweets from Twitter, small(generally textual) units of information will be identified thatremain relatively stable as they spread through the Web. Thetemporal variation within these basic units will be analyzed andmodeled. This modeling will include connections with biologicalmodels of epidemics, as well as new frameworks that exploit thefundamental differences between biological and social contagion.Finally, the temporal variation will be related to network-levelmodels for the diffusion of this information. Generally, the actualnetworks on which real-time information spreads cannot be directlyobserved, nor can the influence of any particular node in the networkbe directly measured. Therefore, the project will develop machine-learning techniques that infer these hidden networks and unobservedlevels of influence.For more information see the project web site at:http://snap.stanford.edu/proj/mipro