LANN: an Integrated Online Annotation Tool for Information Extraction

The creation of high-quality annotated corpora with strong fidelity to the underlying concepts in a domain is a critical step in the development of machine- and deep-learning models for Information Extraction (IE) and Natural Language Processing (NLP). Often, these annotations are done by subject matter experts or highly trained annotators. To achieve high quality suitable for use as a gold-standard corpus, it is necessary that these annotators use common definitions for named entities and relationships, and that they apply them in a uniform manner. In this paper, we present our work on LANN, a text annotation tool that provides support for team-based annotation, quality controls and machine learning assistance across the workflow of an annotation project.