Surveillance, Analysis and Modeling of Chatroom Communities

本提案的目的是开发聊天室通信的信息收集、分析和建模的新技术。首先，研究者和他的同事考虑无图模型来捕捉聊天室通信的结构。特别是，研究人员研究了如何开发一种多维奇异值分解方法，用于聊天室通信数据的成分分析。其次，研究人员开发了新的可视化技术来显示第一步中发现的结构信息。互联网聊天室为不同目标的参与者提供了一个互动的公共交流论坛。聊天室的两个特性使其特别容易被恶意方利用。首先，参与者的真实身份与他们的聊天室昵称分离。第二，多个通信线程可以同时存在。虽然人工监控每个聊天室以确定“谁在和谁聊天”是可能的，但这非常耗时，因此不可扩展。因此，在互联网聊天室中隐藏恶意行为并利用它们进行秘密通信是非常容易的（例如，对手利用青少年聊天室计划恐怖行为）。本项目旨在建立一个完全自动化的监控系统，用于互联网聊天室的数据收集和分析，以发现隐藏的群体。监视是以统计分析的形式对一个特定的聊天、一组聊天或整个聊天室进行的。统计资料用于设计算法，以确定聊天者及其合作伙伴，并回答查询，包括(i)“在哪个聊天室讨论了话题A”，（ii）“谁在聊天室X中讨论话题A”，（iii）“话题A在聊天室X中是热门话题”等。因此，所提出的系统可以帮助情报界在没有人为干预的情况下发现聊天室中隐藏的社区和通信模式。该奖项由美国国家科学基金会和情报界共同支持。数学和物理科学理事会的恐怖主义方法项目支持基础研究和劳动力发展方面的新概念，这些新概念有可能为国家安全做出贡献。

The aim of this proposal is to develop new techniques for information gathering, analysis and modeling of chatroom communications. First, the investigator and his colleague consider graph-less models to capture the structure of chatroom communications. In particular, the investigators study how to develop a multidimensional singular value decomposition approach for component analysis of chatroom communication data. Second, the investigators develop new visualisation techniques to display the structural information found in the first step.Internet chatrooms provide an interactive and public forum of communication for participants with diverse objectives. Two properties of chatrooms make them particularly vulnerable for exploitation by malicious parties. First, the real identities of the participants are decoupled from their chatroom nicknames. Second, multiple threads of communication can co-exist concurrently. Although human-monitoring of each chatroom to determine "who-is-chatting-with-whom" is possible, it is very time consuming, hence not scalable. Thus, it is very easy to conceal malicious behavior in Internet chatrooms and use them for covert communications (e.g., adversary using a teenager chatroom to plan a terrorist act). This project aims at a fully automated surveillance system for data collection and analysis in Internet chatrooms to discover hidden groups. The surveillance is done in the form of statistical profiling for a particular chatter, a group of chatters, or for the entire chatroom. The statistical profiles are used to devise algorithms to determine chatters and their partners and answer to queries including (i) "in which chatrooms topic A is discussed", (ii) "who is chatting about topic A in chatroom X", (iii) "is topic A is a hot one in chatroom X" etc. Thus, the proposed system could aid the intelligence community to discover hidden communities and communication patterns in chatrooms without human intervention.This award is supported jointly by the NSF and the Intelligence Community. The Approaches to Terrorism program in the Directorate for Mathematics and Physical Sciences supports new concepts in basic research and workforce development with the potential to contribute to national security.