This is the first release of the humannotator. You can also find it on [PyPi](https://pypi.org/project/humannotator/0.0.1/) and [Anaconda cloud](https://anaconda.org/lcvriend/humannotator).
The basic idea is implemented: an **annotation tool** that integrates nicely into Jupyter (but plain text display is also supported) and is easy to set up and customize. The annotator is built on top of [pandas](https://pandas.pydata.org/); all annotations are conveniently stored in a dataframe. The user interface is currently mostly based around Python's `input` function. In a future release a more powerful user interface for Jupyter could be created using javascript.
Annotation tasks
Currently the following kind of tasks are supported:
- Strings
- Regexes
- Categories
- Booleans
- Integers
- Floats
- Dates
Some useful improvements which are currently not implemented:
- Selecting spans within a text
- Selecting multiple categories within one task
Data types
Currently the only data type that is supported is the pandas DataFrame and any types of data that can be converted by pandas into a dataframe. The `load_data` function will help convert lists, dictionaries and series into dataframes. Unrecognized data types will need to be converted into a dataframe by the user beforehand.
Support to load text files directly from disk could be useful, as well as supporting other data types such as [spaCy](https://spacy.io/) Docs.
That is all for now.