Named Entity Extraction in a Military Context

摘要

This paper gives an overview of the Named Entity (NE) Extraction experiments performed on the data sets of two cases: (1) Open-Source Intelligence, (2) Post-(Training-)Mission Analytics. In both cases the “tagging” of entities was done as part of wider analytics on the data sets, in a military context. In this context, recognition of military jargon, acronyms and terminology are important and the extraction results were tested for applicability in this area. KNIME Analytics Platform was used for its available models, algorithms and to reduce the effort needed for implementation. Techniques used include NE Model-based Tagging (machine learning) and NE Dictionary-based Tagging. The experiments showed us that the structure of the data set determines which approaches can be successful.