PrettyTags: An Open-Source Tool for Easy and Customizable Textual MultiLevel Semantic Annotations

Labeled data are required for feeding machine learning algorithms and training effectively performing models. Handcrafted annotations of data, made by human experts, require much effort and this task is made heavier when some comfortable tools, for making annotations over the objects, are not available or easily accessible. Furthermore, annotations should be provided in machine-readable formats, to be ready to use in machine learning tasks. In this work, we introduce PrettyTags, an easy-to-use and customizable tool for making text spans annotations, that will be released as an open-source web application. We present a detailed overview of the main features offered by PrettyTags and we also discuss the possibility to link entities annotations in the textual documents to an ontology-based system, for enriching entities semantic representations.