A Semantic Search and Predictive Analytical Framework for Improved Air Cargo Screening

The ability to navigate complex, dense X-ray images with occluded objects and superposition of potential threats presents a significant challenge for air cargo screeners.Explosive detection and threat interdiction is also challenging due to the varied ways threat actors conceal and disguise illicit activity.These factors work to reduce screening efficiency and throughput, as skids must be broken down for secondary or alternative inspection.This proposal outlines a novel Semantic Search and Predictive Analytical Framework to help screeners develop a better understanding of image contents, using manifest/waybill contents to provide visual cues and to make informed predictions about X-ray processing bottlenecks.This approach uses Natural Language Processing (NLP) and artificial intelligence/machine learning (AI/ML) techniques to perform an automated analysis and correlation of cargo contents to the manifest data.Although AI/ML is employed in the proposed system, complex judgement and decision making is dispatched to human screeners.The AI and neural network systems work to surface analytical details and provide visual cues to the screener to facilitate improved detection, interdiction, and efficiency.The approach described in this proposal is based on an open architecture and open data standards to maximize interoperability with different Original Equipment Manufacturers (OEMs) and solution providers.The proposed innovation has promising potential commercial applications.Public and private enterprises perform air cargo screening and would be able to license the software to perform air cargo screening more efficiently and with advanced tools to detect threats, hazards, and illicit activity.