ontonotes-5-parsing 0.0.5

Ontonotes-5-parsing: parser of Ontonotes 5.0 to transform this corpus to a simple JSON format.

A simple parser of the famous Ontonotes 5 dataset https://catalog.ldc.upenn.edu/LDC2013T19

This dataset is very useful for experiments with NER, i.e. Named Entity Recognition. Besides, Ontonotes 5 includes three languages (English, Arabic, and Chinese), and this fact increases interest to use it in experiments with multi-lingual NER. But the source format of Ontonotes 5 is very intricate, in my view. Conformably, the goal of this project is the creation of a special parser to transform Ontonotes 5 into a simple JSON format. In this format, each annotated sentence is represented as a dictionary with five keys: text, morphology, syntax, entities, and language. In their’s turn, morphology, syntax, and entities are specified as dictionaries too, where each dictionary describes labels (part-of-speech labels, syntactical tags, or entity classes) and their bounds in the corresponded text.

You can read more detailed information about this Ontonotes 5 parser in the small documentation https://github.com/nsu-ai/ontonotes-5-parsing/blob/master/readme.md