This tool is designed to analyze and provide insights into the terms and conditions of online services. It uses natural language processing (NLP) to classify sentences within the terms and conditions and provides valuable data insights by detecting potentially unfair clauses.
Demo Link: https://termsnconditionschecker.streamlit.app
-
Enter the Terms and Conditions:
- Paste or type the terms and conditions into the text area provided.
-
Submit:
- Click the "Submit" button to initiate the analysis.
-
Analysis:
- The tool processes the text and classifies sentences based on predefined labels.
- The analysis results include identified labels and sentences for each category.
-
Data Insights:
- Data insights are provided in both bar chart and pie chart formats.
- The charts visually represent the distribution of identified labels within the terms and conditions.
-
Label Descriptions:
- Descriptions of each label used in the analysis are provided for user reference.
- Python 3.x
- Required Python packages are listed in
requirements.txt
.
- The model is finetuned from "legal-bert-base-uncased," a custom BERT architecture designed for the legal domain.
- The model is trained on a dataset of 9.41k sentences for multi-class classification task dataset consisting of terms and conditions of popular online services with 8 classes.
- The model is hosted on Hugging Face and accessed via inference API.
- The app configuration is defined in
app_config.py
. - API configuration is in
api_config.py
. - Classification functions are implemented in
classification.py
. - Label configurations are defined in
label_config.py
.
- Install the required dependencies:
pip install -r requirements.txt
- Run the app:
streamlit run app.py