Semantic Analysis of Military Relevant Texts for Intelligence Purposes

The current deployments of the German Federal Armed Forces cause the necessity to analyze large quantities of intelligence reports and other documents written in different languages. To efficiently handle these tasks natural language processing techniques (NLP) can be applied. The ZENON project makes use of an information extraction approach for the (partial) content analysis of English HUMINT reports. It has further been extended to do multilingual information extraction, i.e., processing Dari and Tajik texts. The focus of this paper is on the improvement of ZENON’s English semantic analysis. Intelligence reports are characterized by a large topical and linguistic variety. In order to extend the system’s coverage when performing content analysis we realized a semantic role labeling approach. In this paper, after a short introduction, the ZENON system and its information extraction functionalities are explained. Then our semantic role labeling approach and the architecture of the implemented application are described in detail.