Aarononomous / ptr

Parse Tree Readability: a new readability metric based on sentence structure

Parse Tree Readability

ptr.py is a python program that measures the readability of a text by inspecting its parse tree, that is, how complicated the sentences are regardless of the words (or word lengths) within them.

Parse Tree Readability.ipynb is a Jupyter notebook that walks through the ptr code. It contains the same functions, etc., presented in a conversational and exploratory way.

There are no other files needed. Libraries such as nltk and spaCy themselves contain corpora, so we can use those (at least to begin with).

Example: Finding the reading level of a text file within the console

>>> from ptr import *
>>> file = open('moby-dick.txt')
>>> text = file.read()
>>> file.close()
>>> corpus = tag_sentences(text)
>>> text_stats = TextStats(corpus).basic_counts()
>>> tag_stats = TagStats(corpus).get_stats()
>>> tree_stats = TreeStats(corpus).get_stats()
>>> readability = ptr(text_stats, tag_stats, tree_stats)
>>> print(readability)
16.74
>>>

Using Parse Tree Readability.ipynb

The Jupyter notebook Parse Tree Readability.ipynb is the best introduction to the theory and usage of this library. Open it (on the command line, with jupyter notebook "Parse Tree Readability.ipynb") and begin exploring.

Installation

Installation is simple. Download (or clone) this repository to your working directory. Then import ptr inside your python file.

Installing the required libraries can be done with pip or conda, your preference, on the command line, ex.:

$ pip install spacy

You may need to download the spaCy english language parser; an error such as Warning: no model found for 'en' will alert you of this. This can be done on the command line as well:

$ python -m spacy download en

License

Licensed under the MIT License. See LICENSE.