Practical overview
When it fits
Use Easse when you have sentence simplification outputs to score against a built-in test set or custom source and reference sentences, or when you need a report to review metrics and examples.
Who it is for
- Researchers comparing sentence simplification systems
- Developers evaluating simplification outputs against reference data
- Teams preparing metric-based reports and example-level reviews
Documented uses
- Compute automatic scores for sentence simplification system outputs.
- Compare systems using commonly used evaluation sets and published outputs.
- Create HTML reports with metrics, plots and samples.
- Analyze word-level transformations or request referenceless quality estimation.
Requirements & limits
The public repository specifies a GPL-3.0 license. The documented installation requires Python 3.6 or 3.7.
- The CLI requires selection of a test set; custom evaluation uses paths for source, predictions and references.
- Available metric selection includes BLEU, SARI, SAMSA and FKGL; SAMSA is disabled by default for speed.
- The project describes its SARI implementation as fixing normalization inconsistencies and an integer overflow issue in the original implementation.
- The documented Python installation uses an editable install from a local repository clone.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Install Easse from its repository
Make the Easse command-line tool available using the documented source installation.
Before you start
- Python 3.6 or 3.7
- Git and pip
Steps
Clone https://github.com/feralvam/easse.git.
Change into the cloned easse directory.
Run pip install -e . from that directory.
Expected outcome
The easse command becomes available and uses the sources in the local repository clone.
Guide limitations
- This covers installation only; it does not evaluate a system output.
- The documented Python requirements are limited to versions 3.6 and 3.7.
Evaluate outputs on a built-in test set
Score a simplification system’s outputs against the TurkCorpus test set using selected automatic metrics.
Before you start
- Easse installed from its repository
- System prediction text available as a file or on standard input
- TurkCorpus test set
Steps
Select the system predictions and the TurkCorpus test set.
Run the evaluate command with the test-set selection and chosen metrics; the README example uses BLEU, SARI and FKGL.
Optionally request referenceless quality estimation or word-level transformation analysis using the documented flags.
Expected outcome
Easse computes the requested automatic evaluation metrics and can include the requested analysis.
Guide limitations
- SAMSA is disabled by default for speed.
- This recipe follows the README’s example; it does not establish results for a particular system.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 16 October 2026
- Original website title
- feralvam/easse · Official homepage title · retained in its source language