A set of scripts for working with PDF files after OCR. They help to extract the plaintext data from the PDFs, to examine a frequency list of words/tokens, to clean the output with regex, to perform kwic, to evaluate CER.
For R scripts: use ocr-quality.Rproj to get already a working directory.
Extract the plaintext from a set of already searchable PDFs.
Clean the plaintext with regex.
Simple code to use regex within R.
A table of word frequencies from the extracted text.
Another exploration of tokens/words frequencies and kwic (keyword in context).
Plaintext file into a simple XML-TEI file with metadata from the filename.
Evaluate CER (Character Error Rate) using scripts in R and Python (Jupyter Notebook).
Usage: load the ground truth text (proofread, checked,...) as a "reference" variable and the text passed by OCR into the "hypothesis" variable.