Demonstrating ConstruKT, a text annotation toolkit for generalized linguistic contructions applied to communication spin

We present ConstruKT, an open and freely available graphical user interface for linguistic annotation of generalized constructions,  which are sets of arbitrary constraints on possibly discontinuous multi-word units and relations between multi-word units or other relations. ConstruKT was written in Python with the TkInter graphical library in the constext of the MiRoR European project, in order to provide a flexible development NLP workbench for research on automatic communication spin detection in research publication for the domain of the health and bio-medical research. The design aims for ConstruKT were to provide a graphical annotation interface for medical domain experts with no expertise in linguistics or Natural Language Processing (NLP), enabling them to annotate arbitrary relations between overlapping discontinuous mutli-word units in a scientific article, and at the same time to have for the project a versatile light weight NLP development workbench in Python. ConstruKT is easily retargetable to any application domain, since its core functionalities define generic multi-word units and binary relations than can be created on the fly and specialized at will.