Crop Circles

The U.S.has no shortage of distinguished colleges and universities. Americans may not be aware, though, of one university’s counter to the post-9/11 War on Terror that is uniquely patriotic and valuable. For years now, expert computer scientists at the University of Arizona’ artificial intelligence lab have been working with focused dedication to accumulate and analyze terrorist Internet content, their response to the ubiquitous anxiety that the United States does not lose the War on Terror on the Web. Their efforts have resulted in the creation of an exclusive resource available to all approved terror-fighting bodies, the Dark Web Project. In their work on the Dark Web Project, the individual University of Arizona researchers have raised their team to the status of being pre-eminent authorities in the methods and technology available for uncovering and tracking terrorists where they foment their violence – in cyberspace.

Terrorists’ use of the Net to foment, recruit, and finance their violence is a growing problem. In 2006, one leading Israeli scholar, Dr. Gabriel Weiman of the University of Haifi, believed there were roughly 5,000 terrorist websites. However, in 2007, researchers in the artificial intelligence lab at the University of Arizona revealed that there were likely 10 times that count– 50,000 terrorist web sites – based on five years of their web spiders moving throughout the Internet and bypassing passwords to gather and document the Dark Web terrorist content in the Web 2.0 environment. In 2008, PC Magazine Online reported that terrorists were disseminating directives and tutorials to followers on how to upload videos of successful attacks to Google’s YouTube video sharing site. Indeed, a crucial portion of online terrorist videos concern to IEDs (improvised explosive devices).

Those who know anything at all about terrorist forces in the world understand that the Internet has long been exploited as a recruiting and radicalizing tool as well as a fund-raising tool. Members of the University of Arizona’s Dark Web Project team reported in early 2010 that they had collected more than 1 million still images and 15,000 videos from terrorist sites and saved them to their servers. These files, as well as scripted content by terrorist/extremist authors (e.g. blog posts, forum chatter, and social networking posts) are available to approved users (the military or intelligence community, for example) via a searchable portal or interface.

In creating the Dark Web Project, Dr. Hsinchun Chen and his research team of computer and information scientists at the University of Arizona’s artificial intelligence lab have responded to what they believe is the “dire jeopardy” of the U.S. and its allies losing the War on Terror in cyberspace. Funded by the National Science Foundation and some other government agencies and private sources, the Dark Web Project team utilizes a wide variety of software system tools they have developed or adapted to document and identify sources of terrorist content. Using strategies within the discipline of artificial intelligence, the software programs seek to unmask and make sense of the terrorists’ world for the benefit of intelligence agencies and military leadership.

First and foremost, the Dark Web team uses hacking techniques to send spiders and web crawlers to collect content and decrypt web interactions. In 2010, the Dark Web researchers claimed to periodically collect the complete contents of about 300 terrorist forums, with some large radical sites boasting more than 30,000 members and nearly a million messages. They had also uncovered a terrorist presence in at least 30 virtual world sites. The team focuses on known Al Qaeda sites as well as some home grown terror cells in Europe. They believe they have the largest open-source terrorist aggregation in the world, though(like everyone else) they have no idea what the U.S. intelligence or defense agencies are doing.

These University of Arizona researchers, who describe themselves as more “computationally oriented” than other anti-terrorism strategists and analysts, have invented and implemented a wide array of analytical tools that each contributes to an integrated understanding of how terrorism on the Web is evolving, and yields specific information that can be used by anti-terrorism forces fighting our enemies in the real world. Identities and connections are just a jump point for what the computer scientists are able to discern about the Dark Web.

One forensic linguistics tool they developed, Writeprint, identifies unique authors of terrorist rhetoric and hate speech with 95 percent accuracy, enabling the U.S. and its allies to take action against them. Dr. Chen led the team that developed the Writeprint program, the name alluding to “written fingerprint.” Its sophistication and multi-language capability make it very difficult for terrorists to scam this invention to bypass the anti-terrorism efforts of the U.S. and its allied Western nations. If it were possible to outsmart the Writeprint program, every terrorist would essentially need to be an expert linguist, consciously altering his usual composition pattern and style as he would post his hate speech to forums or blogs in one of the tool’s languages. (But then, wouldn’t the ” altered” text bear its own Writeprint?)

Writeprint identifies unique authors of anonymous content by analyzing four aspects of written communications: lexicon (vocabulary richness); syntax (writing style with respect to punctuation and function words); structural features (sentence duration, paragraph length, capitalization, and style rules); and content features (keywords and themes). The software establishes and codes an author’s background characteristics from an initial sample, and then is able to actively detect similar background characteristics while processing batches of written material, further refining the unique author’s profile with a high degree of accuracy (95 percent). The program will also detect and integrate strange idiosyncrasies of an author if any are present in the writing sample. In civilian or peacetime applications, Writeprint has also been used to determine the true authorship of some great works of literature and historical documents.

In addition to Writeprint, the Dark Web Project team uses their computer software to analyze the structure of social networks, to distinguish the degrees of connections between terrorist web sites and forum postings, and to visualize or map their relationships. The research team also analyzes and codifies the web site or web page content by selected categories (such as enlisting, training, propaganda, etc.). Another aspect of analysis pertains to the technical sophistication of each web site, measuring attributes such as media richness, interactivity, technical capabilities (e.g., with respect to forms, tables, and multimedia files). The point of the web metrics study is to judge and track the “web savvy-ness” of each of the Dark Web site owners and to integrate these findings with other results to arrive at a more complete profile.

Since not all sites are equally radical, the Dark Web researchers have designed an analysis tool to measure a site contents’ message sentiments and affect to offer insight into how infectious the content is and how much of a radicalizing consequence it might produce.

The Dark Web team also focuses on analyzing the videos that are collected. They have found that a large portion of the videos initiate from a just a small number of sites. The researchers who have analyzed the IED videos particularly have created a organization of unique signatures for the videos with respect to their sources. According to the Dark Web team, the main purpose of collecting the training videos and the violent IED videos is the hope that the U.S. military is analyzing them further to implement findings for the benefit of our own troops and their training prior to being deployed.

Artificial intelligence, as a discipline within the computer sciences, seeks to advance the integration of analyses to mimic assorted aspects of human intelligence, such as language processing, social intelligence, and creativity. The Dark Web Project researchers have not only created the software (spiders) to hack in and collect the content of our enemies’ password-protected web sites and email, they have also implemented a collection of analytical tools to discern the structures, groups, and subgroups of the enemies of the United States from the content they’ve collected. Some of these new inventions, such as Writeprint, have enabled the Dark Web team and other users of the Dark Web Project to identify and track unique authors who are influential and/or who are active terrorist leaders.

The researchers’ goal is to provide information and insight to the soldiers on the front lines as well as to the agents who have infiltrated the enemies’ ranks. In doing so, they have also generated significant academic findings. In addition to creating a unique resource with ingenious analytical software, Dr. Chen and other Dark Web computer scientists at the University of Arizona’s artificial intelligence lab are committed to the ongoing training of researchers who will be prepared to win the War on Terror’s battle in cyberspace for years to come.