Explainable spatiotemporal reasoning for geospatial intelligence applications. Issue 6 (4th May 2022)

• Record Type: Journal Article
• Authors: Duckham, Matt; Gabela, Jelena; Kealy, Allison; Khan, Mustafizur; Legg, Jonathan; Moran, Bill; Rumi, Shakila Khan; Salim, Flora D.; Sharmeen, Shaila; Tao, Yaguang; Trentelman, Kerry; Vasardani, Maria
• Is Part Of: Transactions in GIS. Volume 26:Issue 6(2022)
• Page Start: 2455
• Page End: 2479
• Publication Date: 2022-05-04
• Subjects: Geographic information systems -- Periodicals; 910.285
• DOI: 10.1111/tgis.12939
• Languages: English
• ISSNs: 1361-1682

Abstract

Abstract: This article explores the design, implementation, and querying of a prototype system for automated spatial reasoning for geospatial intelligence applications, called NEXUS. The system combines multiple different reasoning components that can support a wide range of spatiotemporal queries. Fundamental to the requirements of intelligence analysts is the need to provide explanations of system outputs that help better inform users and engender trust in the system. The NEXUS architecture leverages semantic web technologies, and in particular the Simple Event Model and PROV‐O ontologies, to support queries not only about reasoner inferences, but also explanations as to why the system arrived at a particular conclusion. The manipulation of location reports in marine automatic identification system (AIS) data is used as a running example to demonstrate the approach. The range of queries developed illustrates both the detection of suspicious activities and explanations of the inference processes used to identify those activities.