py-rouge
py-rouge copied to clipboard
Diego999
Full Python implementation of the ROUGE metric, producing same results as in the official perl implementation.
py-rouge [depends](https://github.com/Diego999/py-rouge/blob/master/rouge/rouge.py#L1) at least on `nltk` and - if I understand correctly - [also](https://github.com/Diego999/py-rouge/blob/master/rouge/tmp/berkeley_db_to_txt.py#L1) on `bsddb3` - these should be specified as dependencies _somewhere_ (e.g. in `setup.py` or `pyproject.toml` or...
There's version 1.1 on [PyPI](https://pypi.org/project/py-rouge/), but no [tags](https://github.com/Diego999/py-rouge/tags) in this repo. Would be nice to have a tagged commit for building from source, so that the test suite (which isn't...
score = Rouge._compute_p_r_f_score(total_hypothesis_ngrams_count, total_reference_ngrams_count, total_ngrams_overlapping_count, self.alpha, self.weight_factor if use_w else 1.0) may change to: score = Rouge._compute_p_r_f_score(hypothesis_count, reference_count, overlapping_ngrams, self.alpha, self.weight_factor if use_w else 1.0) Since "total_hypothesis_ngrams_count, total_reference_ngrams_count, total_ngrams_overlapping_count" have...
Hi Diego, Is there a way we can compute the error for the rouge metric or the confidence interval from the package? Thanks!
The py-rouge failing during evaluation with best. The system generated file has 20K and reference file has 20k german lines. Trace: Evaluation with Best Traceback (most recent call last): File...
