Homeland Security as Catalyst

Innovative software firms are answering the call from U.S. government agencies for advanced analytics to help combat terrorism and criminal activity. What's the potential of this software for strategic business applications?

A variety of business intelligence systems could benefit from such a powerful, free-form association system. BI systems might use the technology to discover associations between internal products and services with highly lucrative customers or external partners; another application would be to uncover less-than-obvious relationships between the effects of various marketing campaigns. CopLink would naturally be a candidate for an enterprise intrusion and internal fraud detection system. Because the system is designed to consolidate, identify, and share information across boundaries and conduct sophisticated analyses for uncovering hidden relationships and associations, it would naturally fall into the category of software with great potential for BI systems.

Hunting for Big Name

LAS NameHunter does what it calls "computational linguistics." The software helps government agencies search and perform precise recognition of foreign names. LAS has compiled nearly a billion multicultural names in its data archive to assist law enforcement, U.S. intelligence, and homeland security organizations in perfecting "watch lists."

What is NameHunter's potential for business applications? It's easy to imagine how this technology might improve marketing efforts by ensuring that you are dealing with the right individual in spite of the multiple or foreign name spellings. The technology could improve IT systems globalization; today, IT must deal with a diversified customer and client base. Using its data archive of names from more than 200 countries, LAS has classified and created statistics about the names for its multilingual recognition software. The company feels that its approach is superior to traditional character-string mechanisms, such as Soundex, which is found in most relational database systems.

NameHunter first looks at a name to determine its name-type or culture of origin. Then, the name search system brings a number of "cultural/linguistic rules" to bear to get the most accurate match.

Through this approach, the software provides an in-depth understanding of names, their formats, gender associations, and likely country of origin. This approach is necessary because many foreign names don't fit the traditional three-part Western format of a given first name, middle name, and surname.

Some cultures also offer different spellings of names that identify the same individual. This naming complexity is something that software for terrorist watch lists must handle; homeland security requires the management of information about people from all regions of the world. NameHunter tackles this issue by providing ranked results based on linguistic and cultural variation patterns (see Figure 3).

FIGURE 3 - Name recognition software can match varied spellings (or misspellings) of the same person's name.

NameHunter and other name recognition products using a similar approach could help businesses with customer service by improving knowledge of cultural and gender variations in customer names. Such software could also improve enterprise data management approaches to names because the system understands multiple spellings (across multiple databases) of the same customer's name.

Adaptive Profiling

Intelligence for homeland security often arises after targeting attention on crimes associated with terrorism, such as identity theft, fraud, and money laundering. One of the most sophisticated money laundering products on the market comes from Searchspace. The company's Sentinel software monitors more than $600 billion a day for such clients as The Bank of New York, Wells Fargo, UBS, and the London Stock Exchange. The software can detect behavior deviation with every single bank or securities customer by comparing their transactions against their "peer group."

Sentinel can pull data from disparate sources together to enable adaptive profiling and analysis using advanced modeling algorithms. Adaptive profiling is effective for discovering suspicious transactions; this method first compares every transaction against its corresponding account history to determine if the behavior is unusual. It then compares the transaction against a relevant peer group, which is industry-specific or based on customer type, for a further risk analysis. This process greatly reduces the number of false positives and provides a clearer understanding of actual risk.

Searchspace's adaptive anti-money laundering system monitors transactions volumes and amounts on individual accounts over dates and types. As a data mining system, this technology could also help e-commerce and CRM applications, which need to observe customer behavior to reduce attrition rates and bring down customer "churn."

Discovery, Vegas Style

One of the most unusual integration applications comes from SRD, a Las Vegas firm providing software for the casino gaming industry via the company's name matching "Anonymous Entity Resolution" technology. This application allows investigators to determine whether a terrorist suspect appears in two separate databases, such as a government watch list and a car reservation system. Originally developed for background checks of casino employment applicants, SRD's software not only sifts through entity identification information like names, phone numbers, and addresses in separate databases: More importantly, it scrambles and encrypts the information using a "one-way hash function." This method converts a name into a character string that serves as a unique identifier, much like a fingerprint. SRD's Non-Obvious Relationship Awareness (NORA) software can dynamically identify correlations using the identifiers across two separate databases.

Organizations frequently have multiple records in their systems that offer slight variations in name, address, phone number, or other differentiating items, even though these records pertain to the same person or organization. Entity resolution software such as NORA can determine whether multiple records that appear to describe different people or organizations are actually records for a single person or organization. Once the system determines that two or more people or organizations are actually the same, the entity resolution software provides a comprehensive, unified view of all available information about that person or organization.

SRD's software is already being applied to business applications, where it improves the quality of the organization's internal customer data by identifying redundant records for the same person. Other applications of SRD's approach would be background checks and fraud detection.

Security as Catalyst

The products I've mentioned here are worth inspection because they could boost your firm's bottom line through new levels of efficiency, intelligence, precision, security, and ROI. The urgency of homeland security is propelling rapid progress in text and data analytics. The U.S. government is devoting enormous funds to homeland security, as it did toward other intelligence and defense agencies during the Cold War. And as it did in past decades, the government's funding and close attention will prove to be a catalyst for new software that intelligent enterprises will want to evaluate for business applications.

Jesus Mena is the author of many articles and books, including Homeland Security Techniques and Technologies (Charles River Media, 2004), Investigative Data Mining for Security and Criminal Detection, WebMining for Profit: E-Business Optimization (Butterworth-Heinemann, 2003) and Data Mining Your Website (Digital Press, 1999).


Knowledge Computing Corp.: