A platform for knowledge assisted conceptual annotation of multilingual texts

This paper presents an experience of creating the PAnT software tool, a platform for conceptual annotation of multilingual texts, a prerequisite for automated processing of unstructured information. The PAnT platform is supported by linguistic knowledge, which allows various levels of annotation. The tool is multifunctional and has flexible settings that ensure its adaptation to multilingual texts of various domains. PAnT can be used both for the automation of the annotation process as such and for linguistic knowledge acquisition and administration. In this paper, the platform and the methodology of its development are presented on the example of its version, tuned for the conceptual annotation of the “Terrorism” domain corpora in Russian, English and French. The use of PAnT as part of a real annotation project is discussed demonstrating a reduction in the total time and human effort necessary for annotating texts up to the gold standard.