Survival Guide: Perspectives from the field

In his days at the Internal Revenue Service, Jesus Mena became an expert in data mining. With the IRS' blessing, he started consulting on the side and even published a book, "Data Mining Your Website," for dot-com companies. While still at the IRS, Mena raised $2 million in venture capital and in 2000 started a company called WebMiner, which helped businesses make sense of customer information. Mena has since sold WebMiner and turned his attention back to the federal government. His goal: Teach agencies how to use data mining to improve homeland security. Mena works from his office in Alameda, Calif., and consults with agencies, including the Air Force and the General Accounting Office, as well as systems integrators such as Northrop Grumman Corp. and Raytheon Co. His 2003 book, "Investigative Data Mining for Security and Criminal Detection," is one of the first data mining books aimed at law enforcement. Mena spoke to Staff Writer Brad Grimes about data mining, his dot-com days and homeland security. You did artificial intelligence work for the IRS. What does the IRS need with artificial intelligence? Back in the late 1980s and early 1990s, the FBI and IRS formed AI labs, primarily to codify their experience into expert systems. At the IRS, they wanted to classify income tax returns to identify which cases to audit. However, expert systems proved to be too expensive to maintain. I went in to the IRS AI lab because I was interested in extracting rules from the data itself using neural networks and machine-learning algorithms. I was the agency's first data miner. You then went into the commercial market and founded WebMiner. How did you like e-business compared to working for the government? You do things at a startup such as enhance the software in days, rather than years. In an e-business, you are driven by the need to generate revenue and sales continuously, or you are gone. Now you've turned your data mining expertise toward homeland security. How are the data mining needs of business and government the same or different? They are really very similar; however, for homeland security the tasks are especially challenging. Data mining has evolved since I first got started 10 years ago. Back then we worked on data marts and warehouses, both with the IRS and in private industry. The data was imported as tables or flat files into the mining tool. Today data mining has evolved into embedded and distributed modes. Data mining operations can be embedded inside databases. However, for homeland security, data mining will have to be distributed, because databases cannot be moved to a centralized warehouse. There are privacy and security obstacles that prohibit moving them. As such, agent technology will need to be incorporated to send data mining algorithms over networks, and mine the data where it is stored. Homeland security will accelerate the development of this technology, which in the end will benefit private industry, because it will be a cheaper, more effective method of obtaining business intelligence without data warehouses.You're holding a Homeland Security Bootcamp. What are three key points you hope attendees come away understanding? First, they should understand the various processes involved in creating a viable homeland security system. Second, that this will require network-centric, standards-based software in an ever-evolving hybrid system. And third, that these solutions do not need to be built from scratch, but that data sources and software already exist.Did you know there are 200 ways "Mohammed" is spelled in our alphabet, but there is only one way in Arabic? Or that every U.S address and name is normalized monthly into a unique 16-digit identification number? And it is not done by the government; it has been done for years in private industry because of marketing needs. The goal here is to make the bootcamp attendees the most knowledgeable people in their organizations on homeland security techniques, technologies and viable data and software solutions. Back when you were doing field investigations for the IRS, what was the most egregious tax fraud you ever uncovered? You finally asked me a question I can't answer. I recall signing something when I left the IRS that prohibits me from ever disclosing anything I ever did while under their service. You know the government. Mena will hold a Homeland Security Bootcamp for government agencies and integrators Feb. 18 and 19 in Alexandria, Va. For details, go to www.prominer.com.