Introduction
The Department of Homeland Security (DHS) is leveraging advanced artificial intelligence to enhance its operations across a range of critical areas. AI is being employed to streamline data management through entity resolution and improve investigative capabilities by integrating global trade data and enhancing maritime detection. Innovations in geospatial imagery and facial recognition further support situational awareness and response efforts, while AI-driven tools for document review, email analytics, and sentiment analysis improve efficiency and accuracy. Additionally, AI is optimizing user interactions through chatbots and refining processes for immigration case predictions and family matching. The following use cases reflect DHS’s commitment to using cutting-edge technology to enhance security, operational efficiency, and public service.
Use Cases
-
1. Data and Entity Resolution
Data and Entity Resolution automates the process of unifying data and resolving entities with high accuracy and efficiency at an enterprise scale. It employs Machine Learning models to ingest data from various sources and develop associations between disparate records, identifying probable connections, unique entities, and commonalities among independently submitted records. This automation is facilitated by a user-friendly tool that allows non-technical users to continuously train the models. This capability significantly enhances data management by ensuring that records are accurately linked and resolved, thereby improving data integrity and usability across the organization. The system’s scalability and speed make it a valuable asset for handling large volumes of data efficiently.
-
2. Entity Resolution
Entity Resolution leverages third-party global trade data to enhance and enrich investigations into entities of interest. By combining data from companies and goods across multiple languages, it provides network analysis to assess trade flows and associated risks in cross-border trade. This capability helps validate existing information held by the agency or offers a deeper understanding of networks of interest, thereby informing cross-border investigations more effectively. The system utilizes AI and Machine Learning models to manage the information, including data collection, structuring, entity resolution, network and risk analysis. These functions contribute to a comprehensive software knowledge graph and user-friendly frontend interface, enabling end users to interact with and analyze the data efficiently. This approach enhances the agency’s ability to conduct thorough and informed investigations.
-
3. Geospatial imagery utilizing annotation
Geospatial imagery utilizing annotation takes advantage of a commercial constellation of Synthetic Aperture Radar (SAR) satellites, which provide data capable of imaging any location on Earth, day or night, regardless of cloud cover. The system employs AI technologies such as machine vision, object detection, object recognition, and annotation to identify airframes, military vehicles, and marine vessels. Additionally, it includes built-in change detection capabilities that are particularly useful for disaster response missions. This technology ensures continuous and reliable monitoring of various objects and environments, enhancing situational awareness and response capabilities in critical situations. The ability to operate under all weather conditions makes it a robust tool for both routine surveillance and emergency scenarios.
-
4. Integrated Digital Environment
The Integrated Digital Environment offers managers a comprehensive view of end-user workflows, highlighting the most and least used applications and identifying opportunities for improvement. By applying AI and Machine Learning models to user activity data, such as application usage and flow between applications, it helps Customs and Border Protection (CBP) optimize the configuration of interfaces, resource utilization, and the development and deployment of applications. The system generates tailored analytics and insights, enabling metrics gathering, usage observation, dashboarding, and workflow experimentation to support analysts. It also customizes existing capabilities to provide the necessary automations for agency applications and systems. This creates a more connected and secure digital environment, enhancing the ability of CBP administrators to manage and optimize application usage effectively. The integrated approach ensures better connectivity and security across various applications, improving overall operational efficiency.
-
5. RVSS Legacy Overhauled System Project (INVNT)
Video Computer Aided Detection (VCAD) is software that allows Customs and Border Protection (CBP) end users to create and share vision detectors. These VCAD detectors are trained computer vision models capable of recognizing objects, people, and events in any image or video stream. Once trained, the detectors can monitor live video streams in real-time or efficiently search through pre-recorded video data or images to identify objects, people, and events of interest. This capability significantly enhances the ability to detect and analyze important visual information, improving situational awareness and response times. The system’s flexibility in handling both live and recorded data makes it a valuable tool for various surveillance and monitoring applications.
-
6. Use of technology to identify proof of life
The technology for identifying proof of life, known as “Liveness Detection,” employs Artificial Intelligence to minimize fraudulent activities, particularly within Customs and Border Protection (CBP) One app. This app serves as a single portal for various CBP services, catering to travelers, importers, brokers, carriers, international organizations, and other entities through a unified login. It uses guided questions to help users find the appropriate services, forms, or applications they need. The Liveness Detection feature utilizes the user’s mobile device camera along with AI algorithms to verify that the face presented to the app is indeed the person in front of the camera, and not a photo, mask, or other spoofing method. Ensuring the authenticity of submitted data is crucial for the app’s functionality within the agency, providing confidence that the individual is who and where they claim to be. This enhances the security and reliability of the app’s services.
-
7. Vessel Detection
Vessel Detection integrates advanced technologies and analytics to enhance maritime detection and the sensor network. By using machine-assisted and AI-enhanced detection and tracking, it improves the identification of illicit vessels in areas with high volumes of legitimate trade and recreational water traffic. This increases situational awareness and responsiveness to potential threats. The system allows agents to define a search area with specific criteria (e.g., people, drones, vehicles) and transmit these criteria to the sensors. Detected images are automatically recognized using AI algorithms, which filter, detect, and classify objects into Items of Interest (IoI) and “other” objects. IoIs are shared with other detection systems, while non-relevant objects (e.g., animals) are not. The system ensures seamless tracking and maintenance of IoIs across multiple sensors, enhancing maritime security and operational efficiency.
-
8. Text Analytics for Survey Responses (TASR)
Text Analytics for Survey Responses (TASR) is an application that uses Natural Language Processing (NLP) and text analytics to analyze survey responses. Currently, the Department of Homeland Security’s Office of the Chief Human Capital Officer (OCHCO) uses TASR to extract significant topics and themes from unstructured text responses to open-ended questions in the quarterly DHS Pulse Surveys. The extracted results are provided to DHS leadership to better inform agency-wide efforts aimed at meeting employees’ basic needs and improving job satisfaction. This application enhances the understanding of employee feedback, enabling more targeted and effective responses to their concerns and needs.
-
9. Relativity One
RelativityOne is a document review platform designed to improve efficiency in reviewing documents for litigation, Freedom of Information Act (FOIA) requests, and other scenarios requiring large-scale document review and production. This platform streamlines the document review process, making it faster and more efficient. By leveraging advanced technologies, RelativityOne helps users manage and analyze large volumes of documents, ensuring thorough and accurate reviews. This capability is particularly valuable in legal and regulatory contexts where timely and precise document handling is critical.
-
10. Normalization Services
Normalization Services provided by Homeland Security Investigations (HSI) use Artificial Intelligence to verify, validate, correct, and normalize various data elements such as addresses, phone numbers, names, and ID numbers. This streamlines the process of correcting data entry errors, identifying purposeful misidentification, and connecting information about individuals across HSI datasets, thereby reducing the resource hours needed for investigations. Examples of these services include normalizing less well-defined addresses into usable formats, inferring ID types based on provided values, categorizing name parts while considering additional factors, and validating and normalizing phone numbers.
-
11. Email Analytics
The Email Analytics application allows users to review and analyze email data obtained through legal processes. It incorporates AI to classify spam messages and perform named entity recognition (NER) to extract names, organizations, locations, and other entities from the emails. Additionally, the application integrates machine translation capabilities using a commercial product, enabling the analysis of emails in multiple languages. This tool enhances the efficiency and accuracy of email data review, making it easier to identify relevant information and patterns. The AI-driven features streamline the process, saving time and resources in legal and investigative contexts.
-
12. Facial Recognition Service
The Facial Recognition Service is utilized by Homeland Security agents and analysts during investigations to identify known individuals and extract faces for further investigation. This service is particularly useful in cases involving child exploitation offenses, human rights atrocities, and war criminals. The project is part of the Department of Homeland Security Innovation Lab’s Repository for Analytics in a Virtualized Environment (RAVEn). RAVEn supports large, complex analytical projects to enforce and investigate violations of U.S. laws. It also provides tools to analyze trends and isolate criminal patterns as needed.
-
13. I-485 Family Matching
I-485 Family Matching is designed to create models that match family members to their underlying I-485 petitions. The underlying immigrant petition indicates whether the I-485 is employment-based or family-based and includes information about visa classification and priority date, which helps predict visa usage when compared to the Department of State’s monthly Visa Bulletin. Matching an I-485 to its underlying immigrant petition is challenging because the only available matching field is the A-number, which is not always present, and name/date of birth matching is unreliable. The goal of I-485 Family Matching is to leverage AI to confidently connect petitioners and their families based on limited data.
-
14. I-539 approval prediction
The I-539 approval prediction project aims to train and build a machine learning throughput analysis model to predict when an I-539 “Application to Extend or Change Nonimmigrant Status” case will be approved through eProcessing. This model seeks to improve the efficiency and accuracy of the approval process via the eProcessing channel. By leveraging machine learning, the project aims to provide more reliable predictions, enhancing the overall processing experience for applicants. This approach can lead to potential improvements in the approval process, making it more streamlined and effective.
-
15. Sentiment Analysis - Surveys
The Sentiment Analysis – Surveys system offers a statistical analysis of quantitative survey results and uses Natural Language Processing (NLP) modeling software to assign sentiments to categories ranging from strongly positive to strongly negative. This system enables survey administrators to extract valuable insights from employee satisfaction surveys, incorporating both quantitative and qualitative data. The capability to analyze sentiments provides a deeper understanding of employee feedback, helping to identify areas of improvement and enhance overall job satisfaction. This tool is currently available on demand, making it a flexible and valuable resource for ongoing employee engagement efforts.
-
16. Chatbot for International Trade Admin
The Chatbot for International Trade Administration (ITA) is integrated into trade.gov to help ITA clients with frequently asked questions, finding information, and recommending events and services. Clients interact with the chatbot by asking questions or responding to prompts. The chatbot searches through ITA content libraries and staff inputs to provide answers and suggestions tailored to the client’s profile, whether they are an exporter, foreign buyer, or investor. This tool enhances user experience by providing quick and personalized assistance.