Inclusively aims at creating deep learning models that can detect and correct non-inclusive language writing.
It leverages an NLP pipeline that includes:
- Detection: It detects non-inclusive language in the text using text classification techniques.
- Rewriting: If some non-inclusive language is detected, a sequence-to-sequence model is used to modify the text and suggest a more inclusive rewriting.
If you find this project useful, please consider citing the papers listed in the References section.
Inclusively uses a two-stage deep learning pipeline designed to promote inclusive writing. The first stage employs a classification model based on transformer architectures to detect non-inclusive sentences. Once detected, the second stage activates a sequence-to-sequence model, which generates inclusive reformulations of the identified text. The platform also integrates feedback from expert users, allowing continuous model refinement through a human-in-the-loop approach. This system enables Inclusively to offer accurate and contextually appropriate rewrites while also promoting self-learning in inclusive writing practices.
For more information, you can check out the paper here.
- 📊 Classification Model: https://huggingface.co/E-MIMIC/inclusively-classification
- 📝 Reformulation Model: https://huggingface.co/E-MIMIC/inclusively-reformulation-it5
This repository also contains the Flask application that powers the Inclusively platform demo. This demo has been accepted at ECML PKDD 2023 - Demo Track.
The platform is composed of 4 main pages:
- Home: The home page of the platform. It contains a brief description of the platform and a link to the demo.
- Testing: The testing page. It contains the demo itself and the possibility to insert the text to be corrected by Inclusively.
- Evaluation: The evaluation page. It allows users to evaluate the correctness of the corrections made by Inclusively. Each user can provide a feedback for each correction at sentence level.
- Explanation: The explanation page. It contains the output of explainability methods applied to the model. It can allow data scientists to understand how the model works and how it makes its decisions.
Home page
Assistant page
Evaluation and Annotation page
Explanation page
- 🇮🇹 Italian
- [WIP] 🇫🇷 French
- [WIP] 🇪🇸 Spanish
If you find this project useful, please consider citing the following papers:
Inclusively: An AI-Based Assistant for Inclusive Writing
@InProceedings{10.1007/978-3-031-43430-3_31,
author="La Quatra, Moreno and Greco, Salvatore and Cagliero, Luca and Cerquitelli, Tania",
title="Inclusively: An AI-Based Assistant for Inclusive Writing",
booktitle="Machine Learning and Knowledge Discovery in Databases: Applied Data Science and Demo Track",
year="2023",
publisher="Springer Nature Switzerland",
address="Cham",
pages="361--365",
isbn="978-3-031-43430-3"
}A paper describing the methodology and the models trained for this project is currently under review and will be added here upon acceptance.