By Declan Trezise
Customs and border security is a massive job, undertaken by a broad array of agencies worldwide. They all work toward a single goal: enabling the efficient cross-border land, sea, and air passage of legitimate travelers and goods, while concurrently stemming the flow of criminals and contraband. Their success relies on the use of artificial intelligence-powered natural language processing technologies (NLP) to match names and find insights in mountains of publicly available information.
These capabilities are increasingly important in an evolving threat landscape. Customs and border security officials today face an increase in global trade, pandemics, new migration patterns, and a rise in drug trafficking. All these issues make efficient and effective customs and border protection procedures more important than ever.
However, customs and border professionals are too often stymied by outdated technologies. They try to conduct high-stakes name matching using suboptimal search platforms. Their access to publicly and commercially available information (PAI/CAI) is either inadequate or nonexistent.
AI-fueled natural language processing technologies are needed to improve operations.
Matching names and gleaning insight
Often, customs and border security agents rely on ineffective full-text search platforms to match names in structured text, such as when comparing names of incoming travelers against watchlists. The name matching capabilities of these search engines lie somewhere between binary match determinations and fuzzy matching — a computing approach that improves upon binary processes by considering degrees of truth. Returning only exact or near-exact matches, search platforms are fuzzy enough for general searches, but not expansive or fast enough for optimized name matching, and adding fuzziness without name specific tuning adds to the burden of false positives that require human intervention.
Many full-text search engines accommodate only a limited number of languages, making it difficult to match translated names, transliterated names, and names rendered in non-Latin scripts. Match/no match processes also fail to spot aliases, nicknames, misspellings, honorifics, or out-of-order names.
Further, many customs and border security name matching technologies rely on manual processes for data entry and verification — processes that take too long and are too error-prone. Due to these inadequacies, customs and border officials miss too many matches, allowing entry of terrorists, criminals, and contraband. Conversely, by returning too many false positives, these systems lead to unnecessary security alerts and inhibit the movement of legitimate travelers and goods.
AI-powered fuzzy name matching can help customs and border security agencies overcome the challenges of matching names in structured text. Automated NLP algorithms use a variety of criteria to quickly, accurately, and intelligently match and disambiguate names of people, organizations, and locations across a broad array of languages, scripts, and databases.
Publicly available information is any data that is freely accessible by the public. This data includes social media posts, news stories and videos, information appearing on web sites, and more. In a world where people spend 27 percent of their time online[1], creating more than 2.5 quintillion bytes of data daily[2], there is an incalculable amount of PAI available for search.
Searching, monitoring, and analyzing PAI in real time can help customs and border security officials better identify potential threats. These officials can use PAI systems to detect and track illegal cross-border activity; monitor the movements of individuals and groups of interest; and aid in real-time threat intelligence and response planning.
These capabilities help border officials better manage both potential and present threats. Consider a State Department official pre-screening travelers for United States visas. That official can use PAI platforms to examine social media posts and other content to determine whether the applicant is in any way associated with a criminal appearing on a watch list. Benefits are even clearer to those charged with preventing present danger. If a PAI system detects someone tweeting, “Just saw a woman abandon a bag @JFK Airport, Gate 8,” it can trigger an alert to airport authorities.
While some customs and border security organizations currently use PAI systems, those systems are often suboptimal. They may only scan a limited number of data sources; serve up poor quality data; and prove incapable of handling the vast amounts of data necessary for true insight.
Improve operations with a combined name matching/PAI solution
To best secure their countries, customs and borders security personnel need best-in-class combined name matching and PAI solutions. But there are a number of these solutions on the market. What should you look for?
- Interoperability: You cannot obtain the benefits of a combined system if your name matching and PAI platforms don’t work well together, and with your existing technologies.
- Ease of deployment: For the easiest deployment possible, containerized delivery should be available. Deployment options should be up to you: cloud-based or on-site, depending on your organization’s needs.
- An array of PAI capabilities: Look for an automated solution that can access all layers of the internet, including the deep and dark web. Choose a PAI platform that includes a large and diverse library of enriched data, originating from a broad array of web sites; commercially available sources; and real-world interactions generated on chats, social media posts, and online comments. For the most up-to-date insight, it should provide persistent search capabilities — or the ability to keep a search operation open whether someone is using it or not, automatically recording updates and changes to the search term.
- Accurate name matching, and more: Look for name matching systems that are both fast and accurate. Your name matching solution should give you a clear match score, and the ability to adjust scoring parameters to meet your needs.
- Language translation: The best PAI and name matching solutions automatically translate content from an array of different languages, helping customs and border security officials to monitor names and online content from across the globe.
- Entity resolution: Your combined PAI and name matching solution should resolve entities, a process that entails examining personal and corporate names appearing in unstructured text, then matching those names to entities appearing in a public knowledge base or the knowledge bases maintained by your organization. This capability helps you distinguish among multiple entities with the same or similar names.
Too many existing name matching and PAI systems inadequately address the modern threat landscape. They are slow, incomplete, inaccurate, and error-prone. AI-powered, interoperable PAI and name matching solutions can help customs and border security officials better protect their nations.
Endnotes
1. Salim, S, Digital Information World, “More than six hours of our day is spent online – Digital 2019 Reports,” accessed May 2023, https://www.digitalinformation...
2. SG Analytics, “2.5 Quintillion Bytes of Data Generated Everyday — Top Data Science Trends 2020,” accessed May 2023, https://us.sganalytics.com/blo...
